sign in sign up
<<
Home , Cardiovascular disease, Vectorcardiography

Evidence Synthesis Number 163

Screening for Cardiovascular Risk With : An Evidence Review for the U.S. Preventive Services Task Force

Pre pare d for: Agency for Healthcare Research and Quality U.S. Department of Health and Human Services 5600 Fishers Lane Rockville, MD 20857 www.ahrq.gov

Contract No. HHSA-290-2015-00011-I, Task Orde r No. 5

Prepared by: RTI International–University of North Carolina at Chapel Hill Evidence-based Practice Center Research Triangle Park, NC

Investigators: Daniel E. Jonas, MD, MPH Shivani Reddy, MD, MSc Jennifer Cook Middleton, PhD Colleen Barclay, MPH Joshua Green, BA Claire Baker Gary Asher, MD, MPH

AHRQ Publication No. 17-05235-EF-1 June 2018 This report is based on research conducted by the RTI International–University of North Carolina at Chapel Hill Evidence-based Practice Center (EPC) under contract to the Agency for Healthcare Research and Quality (AHRQ), Rockville, MD (Contract No. HHSA-290-2015- 00011-I, Task Order No. 5). The findings and conclusions in this document are those of the authors, who are responsible for its contents, and do not necessarily represent the views of AHRQ. Therefore, no statement in this report should be construed as an official position of AHRQ or of the U.S. Department of Health and Human Services.

The information in this report is intended to help health care decisionmakers—patients and clinicians, health system leaders, and policymakers, among others—make well-informed decisions and thereby improve the quality of health care services. This report is not intended to be a substitute for the application of clinical judgment. Anyone who makes decisions concerning the provision of clinical care should consider this report in the same way as any medical reference and in conjunction with all other pertinent information (i.e., in the context of available resources and circumstances presented by individual patients).

This report may be used, in whole or in part, as the basis for development of clinical practice guidelines and other quality enhancement tools, or as a basis for reimbursement and coverage policies. AHRQ or U.S. Department of Health and Human Services endorsement of such derivative products may not be stated or implied.

None of the investigators has any affiliations or financial involvement that conflicts with the material presented in this report.

Acknowledgments

The authors gratefully acknowledge the following individuals for their contributions to this project and deeply appreciate their considerable support, commitment, and contributions: Howard Tracer, MD, AHRQ Medical Officer, and Elisabeth Kato, MD, MRP, former AHRQ Medical Officer; Tracy Wolff, MD, MPH, AHRQ Associate Scientific Director; current and former members of the U.S. Preventive Services Task Force who contributed to topic deliberations; expert reviewers Joy Melnikow, MD, MPH, Amit Shah, MD, Fabrizio Turrini, MD, and Timothy Wilt, MD; federal partner reviewers at the National Institutes of Health; and RTI International–University of North Carolina EPC staff Carol Woodell, BSPH, Christiane Voisin, MSLS, Laurie Leadbetter, MSLS, Sharon Barrell, MA, and Loraine Monroe, publications specialist.

Suggested Citation

Jonas DE, Reddy S, Middleton JC, Barclay C, Green J, Baker C, Asher G. Screening for Cardiovascular Disease Risk With Electrocardiography: An Evidence Review for the U.S. Preventive Services Task Force. Evidence Synthesis No. 163. AHRQ Publication No. 17-05235- EF-1. Rockville, MD: Agency for Healthcare Research and Quality; 2018.

Screening for CVD Ri sk Wi th ECG ii RTI–UNC EPC Structured Abstract

Purpose: To systematically review the evidence on screening asymptomatic adults for cardiovascular disease (CVD) risk using resting or electrocardiography (ECG) for populations and settings relevant to primary care in the United States.

Data Sources: PubMed/MEDLINE, the Cochrane Library, and trial registries through May 30, 2017; reference lists of retrieved articles; outside experts; and reviewers, with surveillance of the literature through April 4, 2018.

Study Selection: Two investigators selected English-language studies using a priori criteria. Eligible studies focused on the use of resting or exercise ECG for adults without symptoms or a diagnosis of CVD. Eligible designs included controlled trials comparing ECG screening with no ECG screening and prospective cohort studies reporting reclassification, calibration, or discriminat ion that compared risk assessment using ECG plus traditional risk factors versus traditional risk factors alone. For harms of ECG, prospective cohort studies, large retrospective cohort studies, and case-control studies were also eligible. For harms from exercise ECG or subsequent procedures/interventions, large registries or multicenter studies without a control group were also eligible.

Data Extraction: One investigator extracted data and a second checked accuracy. Two reviewers independently rated quality for all included studies using predefined criteria.

Data Synthesis: Sixteen studies (77,140 participants) were included. Two randomized, controlled trials (RCTs) (1,151 participants) found no significant improvement in all-cause mortality, cardiovascular-related mortality, myocardial (MI), failure, or for screening with exercise ECG in asymptomatic adults ages 50 to 75 years with compared with no screening. In addition, there was no significant improvement for their primary composite outcomes (hazard ratio [HR] 1.00 [95% confidence interval [CI], 0.59 to 1.71] for all- cause mortality, nonfatal MI, nonfatal stroke, or requiring hospitalization or emergency service intervention, and HR 0.85 [95% CI, 0.39 to 1.84] for nonfatal MI or cardiac ). No controlled trials evaluated screening with resting ECG. Although potential harms of exercise or resting ECG include , acute MI, sudden cardiac death, and harms of subsequent angiography or revascularization procedures after an abnormal test, evidence on their frequency in asymptomatic persons was scant. Evidence from five cohort studies (9,582 participants; mean baseline [FRS] 10.8 to 12.3 in studies reporting it) shows that the addition of exercise ECG abnormalities to traditional CVD risk factors results in small improvements in discrimination (absolute improvement in area under the curve [AUC] or C-statistics 0.02 to 0.03; 95% CIs rarely reported), but it is uncertain whether calibration or appropriate risk classification improves. Evidence from nine cohort studies (66,407 participants; mean baseline risk ranging from low to high across studies) shows that the addition of resting ECG findings to traditional CVD risk factors results in very small or small improvements in discrimination (absolute improvement in AUC or C-statistics 0.001 to 0.05) and improvements for calibration and appropriate risk classification for prediction of multiple outcomes (e.g., all- cause mortality, CVD mortality, CHD events). Total net reclassification improvements (event;

Screening for CVD Ri sk Wi th ECG iii RTI–UNC EPC nonevent) ranged from 3.6 percent (2.7%; 0.6%) to 30 percent (17%; 19%) for studies using Framingham Risk Score (FRS) or pooled cohort equations (PCE) base models (95% CIs were rarely reported).

Limitations: The RCTs that evaluated exercise ECG in asymptomatic diabetic patients did not reach sample size targets and were stopped early because of trouble recruiting; both followed participants for about 3.5 years. For risk prediction with the addition of ECG, evidence was limited by imprecision, quality, and considerable heterogeneity. Consistency of findings for specific risk thresholds is unknown because all studies used different risk categories. About half of the included risk prediction studies did not use the published coefficients of externally validated base models such as FRS or PCE; only one used the PCE as a base model. For risk prediction with resting ECG, it is unclear what proportion of participants was truly asymptomatic because most studies did not report any assessment of symptoms.

Conclusions: The overall strength of evidence was low or insufficient for each of the questions and outcomes evaluated. RCTs of screening with exercise ECG in asymptomatic participants found no improvement in health outcomes despite focusing on higher risk populations with diabetes. For asymptomatic persons without a history of CVD, the harms of exercise or resting ECG can include arrhythmias, acute MI, sudden cardiac death, and harms of subsequent angiography or revascularization procedures after an abnormal test, but the frequency of these harms is uncertain. Evidence on whether the addition of exercise ECG to traditional CVD risk factors results in accurate reclassification is lacking. Cohort studies found that the addition of multiple resting ECG abnormalities to traditional CVD risk factors accurately reclassifies persons, and improves discriminat ion and calibration, but evidence was limited by imprecision, quality, considerable heterogeneity, and inconsistent use of risk thresholds that align with clinical decisions and recommendations.

Screening for CVD Ri sk Wi th ECG iv RTI–UNC EPC Table of Contents

Chapter 1. Introduction...... 1 Scope and Purpose ...... 1 Condition Definition ...... 1 Etiology and Natural History...... 1 Risk Factors...... 2 Prevalence and Burden...... 3 Rationale for Screening and Screening Strategies ...... 3 Treatment Approaches ...... 4 Recommendations and Clinical Practice in the United States ...... 4 Chapter 2. Methods ...... 6 Key Questions and Analytic Framework ...... 6 Data Sources and Searches ...... 6 Study Se lect ion...... 7 Quality Assessment and Data Abstraction...... 8 Data Synthesis and Analysis...... 8 Expert Review and Public Comment ...... 9 USPSTF Involvement ...... 9 Chapter 3. Results ...... 10 Literature Search ...... 10 Results ...... 10 KQ 1a. Does the Addition of Screening With Resting or Exercise ECG Improve Health Outcomes Compared With Traditional CVD Risk Factor Assessment Alone in Asymptomatic Adults? 1b. Does Improvement in Health Outcomes Vary for Subgroups Defined by Baseline CVD Risk, Age, Se x, or Race/Ethnicity? ...... 10 KQ 2. Does the Addition of Screening With Resting or Exercise ECG to Traditional CVD Risk Factor Assessment Accurately Reclassify Persons Into Different Risk Groups or Improve Measures of Calibration and Discrimination? ...... 12 KQ 3a. What Are the Harms of Screening With Resting or Exercise ECG, Including Harms of Subsequent Procedures or Interventions Initiated as a Result of Screening? 3b. Do the Harms of screening Vary for Subgroups Defined by Baseline CVD Risk, Age, Sex, or Race/Ethnicity?...... 21 Chapter 4. Discussion ...... 23 Summary of Evidence ...... 23 Evidence for the Benefits and Harms of Screening With Resting or Exercise ECG ...... 23 Reclassification, Calibration, and Discrimination With the Addition of Resting or Exercise ECG to Traditional CVD Risk Factor Assessment ...... 25 Limitations ...... 27 Future Research Needs...... 28 Conclusion ...... 28

Figures Figure 1. Analytic Framework Figure 2. Summary of Evidence Search and Selection Diagram

Screening for CVD Ri sk Wi th ECG v RTI–UNC EPC Figure 3. Main Results of Included Randomized, Controlled Trials Reporting Health Outcomes (KQ1) Figure 4. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Models Figure 5. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, Systemic Coronary Risk Evaluation, or Conventional Risk Factor Base Models Figure 6. Effect on Reclassification of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models Figure 7. Effect on Reclassification of Adding Resting ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Models

Tables Table 1. Test Performance Measures for Comparing Risk Assessment or Prediction Models Table 2. Characteristics of Included Randomized, Controlled Trials for KQ 1 and KQ 3 Table 3. Characteristics of Included Studies for KQ 2 That Evaluated Exercise ECG, Part 1 Table 4. Characteristics of Included Studies for KQ 2 That Evaluated Exercise ECG, Part 2 Table 5. Results of Included Studies for KQ 2 That Evaluated Exercise ECG Table 6. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 1 Table 7. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 2 Table 8. Summary of Evidence for Screening With ECG Table 9. Snapshot to Assess Net Benefit

Appendixes Appendix A. Additional Background, Summary of Recommendations From Other Groups, and Contextual Questions Appendix B. Detailed Methods Appendix C. Excluded Studies Appendix D. Quality Assessments Appendix E. Additional Tables and Results

Screening for CVD Ri sk Wi th ECG vi RTI–UNC EPC Chapter 1. Introduction

Scope and Purpose

The U.S. Preventive Services Task Force (USPSTF) will use this report to inform an update of its 2012 recommendation on screening asymptomatic adults with electrocardiogram (ECG) for the prediction of cardiovascular disease (CVD) events.1 In 2012, the USPSTF recommended against screening with resting or exercise ECG for the prediction of coronary heart disease (CHD) events in asymptomatic adults at low risk for CHD events (D recommendation). For asymptomatic adults at intermediate or high risk, the USPSTF concluded that evidence was insufficient to assess the balance of benefits and harms of screening (I statement). The purpose of this report is to systematically evaluate the current evidence on using resting or exercise ECG to screen asymptomatic adults for CVD risk for populations and settings relevant to primary care in the United States. This report summarizes the evidence on the benefits and harms of adding screening with resting or exercise ECG to traditional CVD risk factor assessments compared with using traditional CVD risk factor assessments alone. This report also summarizes the evidence on whether the addition of ECG accurately reclassifies persons into different risk groups or improves measures of calibration and discrimination.

Condition Definition

CVD is a broad term encompassing atherosclerotic conditions that affect the heart and blood vessels.2-4 CVD generally refers to , including but not limited to CHD (also called ischemic heart disease), , and peripheral disease (PAD). In patients with CVD, plaques form within the , causing reduced blood flow and/or arterial blockage. Serious CVD events (sometimes described as hard outcomes) include (MI), heart failure, stroke, and sudden cardiac death. Less severe CVD events (sometimes called soft outcomes) include , , transient ischemic attack (TIA), and revascularization.

Etiology and Natural History

In CVD, atherosclerotic plaque is deposited over many years within the endothelial lining of the , which provide oxygenated blood to the myocardium. The development of the plaque, containing not only but other molecules secreted from various types of cells, is induced by a cascade of mechanisms including inflammatory processes.5, 6 Some sites, such as branch points and the inner curves of arteries, are more susceptible to deposition of atherosclerotic plaque.7 Sudden plaque rupture, or intra-luminal related to the exposure of the ruptured plaque’s thrombogenic core, is associated (though not invariably) with , MI, and sudden cardiac death.7, 8 Progression of atherosclerosis and of CVD is influenced by a variety of risk factors, some of them modifiable and thus targets for intervention.

Screening for CVD Ri sk Wi th ECG 1 RTI–UNC EPC Evidence of obstructive CHD, upon evaluation by coronary angiography, has been considered to be prognostic of significant morbidity and mortality; however, it has been demonstrated that culprit lesions, in patients who experience acute coronary events due to plaque rupture or acute thrombosis, may not be those angiographically observed to significantly occlude the vessel.8, 9 Some persons do not experience symptoms before major first CVD events10 because major first events can result from plaque rupture in vessels without significant .

Risk Factors

Traditional risk factors for CVD are male sex, older age, smoking, , (high total or low-density lipoprotein or low high-density lipoprotein [HDL] cholesterol), and diabetes. They are independently associated with risk of CVD and are included in the traditional Framingham risk assessment model.11, 12 Some risk factors are modifiable and could be targets for treatment in patients identified as being at higher risk. Prevalence of risk factors is high in the United States: as of 2014, 16.9 percent of U.S. adults smoked ; as of 2012, 69 percent were or obese, 13.1 percent had serum total cholesterol levels ≥240 mg/dL, 32.6 percent were hypertensive, and 8.5 percent of adults had diagnosed diabetes mellitus (another 3.3% had undiagnosed diabetes mellitus).13 More than 90 percent of CVD events occur in persons with one or more risk factors.14 Other factors, some behavioral and others biomarkers, are not included among the major independent traditional risk factors. Although they increase CVD risk to varying degrees, they have not been shown to consistently improve risk prediction in a clinically meaningful way beyond traditional risk factors; these include family history of early CVD, , physical inactivity, atherogenic , and presence of prothrombic and proinflammatory factors.12, 13, 15

To help providers operationalize the large number of factors that need to be considered, risk prediction equations that integrate and weight the traditional risk factors are used commonly in clinical practice to assess 10-year risk of CVD events and to guide treatment decisions. The USPSTF recommends using the American College of /American Heart Association (ACC/AHA) pooled cohort equations (PCE) to calculate 10-year risk for adults ages 40 to 75 years to inform decisions about use for prevention of CVD and to inform decisions about use for primary prevention.2, 16, 17 The USPSTF noted that concerns have been raised about the PCE’s potential to overpredict risk and their moderate discrimination, but also that they are the only U.S.-based, externally validated equations that report risk as a combination of cerebrovascular and CHD events.2, 17 The PCE approach takes into account age, sex, race, cholesterol levels, systolic level, antihypertension treatment, presence of diabetes, and smoking status as risk factors and focuses on prediction of hard outcomes, specifically, heart attack and death from CHD, ischemic stroke, and stroke-related death.16 Using the PCE and National Health and Examination Survey data from 2011–2012, an estimated 9.4 percent of adults ages 40 to 79 years without a history of CVD have a 10-year risk greater than 20 percent.18 Age has a large influence on the PCE’s predicted risk, and it is estimated that 41 percent of men and 27 percent of women ages 60 to 69 years without a history of CVD will have a 10-year risk of at least 10 percent.19

Screening for CVD Ri sk Wi th ECG 2 RTI–UNC EPC Prevalence and Burden

CVD is the leading in U.S. adults and causes about a third of all .13, 18 In the United States, an estimated 580,000 persons have a first MI each year and about 610,000 have a first stroke.20 Adult CVD prevalence increases with each decade of life, with higher prevalence among men than women.13 The average annual incidence of first major cardiovascular (CV) event increases from around 25 cases per 1,000 in men ages 35 to 44 years to 80 cases per 1,000 in men age 85 years or older. For women (compared with men), similar incidence rates are observed about 10 years later in life, although the gap narrows for women ages 75 to 84 years and is reversed by age 85 years or older. Prevalence of CHD and stroke were nearly 3 to 4 times greater for adults age 65 years or older than for those ages 45 to 64 years (19.8% vs. 7.1% and 8.3% vs. 2.9%, respectively).21, 22 Disparities exist with regard to mortality from and prevalence of CVD. Mortality rates are lowest for white women and highest for black men, and prevalence is highest for American Indians/Alaska Natives and blacks.21, 22 CVD is a major source of direct and indirect health care costs in the United States. The estimated total cost of CVD in 2015 was $182 billion, which is predicted to double by 2030.13

Rationale for Screening and Screening Strategies

Because many patients do not have any symptoms of CVD before a serious first event, such as MI or stroke, identifying asymptomatic individuals for treatment with preventive medications may reduce risk for future CVD morbidity and mortality. Approximately 30 percent of patients presenting with acute coronary syndromes do not have a prior diagnosis of CVD.10, 23 For screening with resting or exercise ECG to be effective, it must be able to reclassify individuals in a manner that results in treatment changes that improve health outcomes. For example, screening with ECG might reclassify persons into higher or lower risk categories. Appropriately reclassifying such individuals could help target use of preventive interventions to those most likely to benefit or could reduce use of preventive interventions for those least likely to benefit. However, if reclassification was inappropriate, it could to an increase in overtreatment or undertreatment.

Potential screening strategies include both resting and exercise ECG. Resting ECG records cardiac electrical activity over a short time, typically 10 seconds. Clinicians interpret the recorded ECG waveforms to look for evidence of conduction problems and/or myocardial . Resting or exercise ECGs have long been used as tests for the diagnostic evaluation of suspected CVD, which has led to consideration of their use for screening asymptomatic individuals and risk prediction. Although the most common method of exercise testing is the exercise treadmill test, other methods include bicycles and ergometers. Both resting and exercise ECG may show markers of unrecognized previous MI, silent or inducible myocardial ischemia, and other cardiac abnormalities (such as left , , or arrhythmias) that may be associated with CVD or may predict future CVD events.

Screening for CVD Ri sk Wi th ECG 3 RTI–UNC EPC Treatment Approaches

If screening for CVD risk results in appropriate risk reclassification, intensified preventive interventions focus on -lowering therapy and aspirin, which have been demonstrated to reduce risk as assessed by a variety of outcomes.24-28 Other preventive interventions (, blood pressure control, and ) would not be significantly affected by reclassification of risk (e.g., recommendations for smoking cessation are the same regardless of risk classification). Current USPSTF recommendations for and aspirin for primary prevention are based on the 10-year CVD risk as estimated by the PCE.2, 17 The USPSTF recommends initiating use of low- to moderate-dose statins in adults ages 40 to 75 years without a history of CVD who have one or more CVD risk factors (dyslipidemia, diabetes, hypertension, or smoking) and a calculated 10-year CVD risk of 10 percent or greater (B recommendation) and selectively offering statins to adults ages 40 to 75 years without a history of CVD who have one or more CVD risk factors and a calculated 10-year risk of 7.5 percent to 10 percent (C recommendation). For aspirin, the USPSTF recommends initiating low-dose aspirin use for the primary prevention of CVD and colorectal cancer in adults ages 50 to 59 years who have a 10% or greater 10-year CVD risk, are not at increased risk for , have a of at least 10 years, and are willing to take low-dose aspirin daily for at least 10 years (B recommendation). The USPSTF also has a C recommendation for initiating low-dose aspirin use for the primary prevention of CVD and CRC in adults ages 60 to 69 years who have a 10 percent or greater 10-year CVD risk. The ACC and AHA jointly released guidelines (2014) recommending that moderate- to high-intensity statin therapy should be used in persons ages 40 to 75 years without a history of clinical CVD or diabetes and with an estimated 10-year risk of 7.5 percent or greater (Grade A, strong recommendation).29 The ACC/AHA also recommended moderate-intensity statins when risk is 5 percent to less than 7.5 percent (Grade C, weak recommendation).

Recommendations and Clinical Practice in the United States

Numerous organizations recommend against routine screening of asymptomatic adults for CVD with resting or exercise ECG, including the American College of Physicians,30 the American Academy of Family Physicians,31 and the American College of Preventive (Appendix A Table 1).32-34 Screening of special populations is recommended by some groups. For example, the American Academy of Family Physicians recommends screening otherwise low-risk patients who have certain occupations in which undetected CVD could significantly affect the public (e.g., airline pilots),35 and the American College of Sports Medicine recommends screening moderate-risk patients who are beginning a new exercise regimen.36

Many risk prediction equations (i.e., tools, models, scores, calculators) are available and have been recommended in various countries for use in clinical practice to guide treatment decisions (Appendix A Table 2). The USPSTF and the ACC/AHA recommend using the PCE to calculate 10-year risk. The existing equations that have been externally validated vary in the risk factors included and predicted outcomes (e.g., global CVD outcomes vs. mortality vs. CHD-specific outcomes). The Framingham Risk Score (FRS) was the first widely used multivariable risk assessment tool33, 34, 37 and included sex, age, total and high-density lipoprotein cholesterol

Screening for CVD Ri sk Wi th ECG 4 RTI–UNC EPC (HDL-C), blood pressure, diabetes, and smoking. A variety of Framingham-based risk equations have been externally validated.27, 33, 34, 37-39 One early Framingham model (1991)33 included left ventricular hypertrophy (LVH) (determined by ECG) along with traditional risk factors, but it was dropped from later models. None of the currently recommended prediction equations include ECG.

Despite recommendations, use of risk assessment in clinical practice may be suboptimal. For example, a survey of over 900 U.S. physicians found that although more than 80 percent agreed that risk calculation is useful, only 41 percent reported that they use it in practice.40 Among those who use it, the majority use it to guide lipid-lowering therapy recommendations (69%) and aspirin therapy recommendations (54%).40 Limited data are available on using ECGs to assess CVD risk in asymptomatic persons, but a population-based retrospective cohort study of Canadian adults reported that 21.5 percent had an ECG within 30 days of an annual health exam.41 The proportion varied widely across 679 primary care practices (from 1.8% to 76.1%).

Screening for CVD Ri sk Wi th ECG 5 RTI–UNC EPC Chapter 2. Methods

Key Questions and Analytic Framework

The Evidence-based Practice Center (EPC) investigators, USPSTF members, and Agency for Healthcare Research and Quality (AHRQ) Medical Officers developed the scope and key questions (KQs). Figure 1 shows the analytic framework and KQs that guided the review. Three KQs were developed for this review:

1a. Does the addition of screening with resting or exercise electrocardiography (ECG) improve health outcomes compared with traditional CVD risk factor assessment alone in asymptomatic adults? 1b. Does improvement in health outcomes vary for subgroups defined by baseline CVD risk (e.g., low, intermediate, or high risk), age, sex, or race/ethnicity? 2. Does the addition of screening with resting or exercise ECG to traditional CVD risk factor assessment accurately reclassify persons into different risk groups (e.g., high-, intermediate-, and low-risk groups) or improve measures of calibration and discriminat ion? 3a. What are the harms of screening with resting or exercise ECG, including harms of subsequent procedures or interventions initiated as a result of screening? 3b. Do the harms of screening vary for subgroups defined by baseline CVD risk (e.g., low, intermediate, or high risk), age, sex, or race/ethnicity?

In addition to addressing our KQs, evidence related to two Contextual Questions was assessed. The first focused on what medications (i.e., aspirin, lipid-lowering therapy) are recommended for persons in each CVD risk category and the fidelity to prescribing and taking the recommended medications. The second focused on the harms and benefits of revascularization procedures in adults without symptoms or a prior diagnosis of CVD. These Contextual Questions were not a part of the systematic review. They are intended to provide additional background information. Literature addressing these questions is summarized in Appendix A.

Data Sources and Searches

PubMed/MEDLINE and the Cochrane Library were searched for English-language articles published from January 2009 through May 30, 2017. Medical Subject Headings were used as search terms when available and keywords when appropriate, focusing on terms to describe relevant populations, tests, interventions, outcomes, and study designs. The search relied primarily on the 2011 systematic review for the USPSTF42 to identify potentially relevant studies published before 2009 (we reassessed all articles included in that systematic review using the eligibility criteria). Complete search terms and limits are listed in Appendix B1. Targeted searches for unpublished literature were conducted by searching ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry Platform. To supplement electronic searches, the reference lists of pertinent review articles and studies that met the inclusion criteria were reviewed and all previously unidentified relevant articles were added. All

Screening for CVD Ri sk Wi th ECG 6 RTI–UNC EPC literature suggested by peer reviewers or public comment respondents was reviewed for eligibility. Since May 2017, ongoing surveillance was conducted through article alerts and targeted searches of journals to identify major studies published in the interim that may affect the conclusions or understanding of the evidence and the related USPSTF recommendation. The last surveillance was conducted on April 4, 2018 and no additional studies meeting eligibility criteria were identified. All literature search results were managed using EndNoteTM version 7.4 (Thomson Reuters, New York, NY).

Study Selection

Inclusion and exclusion criteria were developed for populations, interventions, comparators, outcomes, settings, and study designs with input from the USPSTF (Appendix B2). English- language studies of adults age 18 years or older without symptoms or a diagnosis of CVD were included. Studies of children, adolescents, and persons with a history of CVD or symptoms suggesting CVD were excluded. Studies assessing resting ECG or exercise ECG were included and studies that assessed radiology tests, , and vectorcardiography were excluded. Eligible studies recruited participants from primary care settings, occupational medicine settings, or the general population in countries categorized as “very high” on the Human Development Index.

For all KQs, controlled clinical trials and randomized, controlled trials (RCTs) comparing groups that were screened with groups that were not screened (i.e., comparing risk stratification using ECG plus traditional risk factors vs. risk stratification using traditional risk factors alone) were eligible. For KQ 1 (direct evidence that screening improves health outcomes), eligible outcomes included all-cause mortality, CV mortality, and CV events (MI, angina, stroke, congestive heart failure, composite CV outcomes).

For KQ 2 (calibration, discriminat ion, and reclassification), prospective cohort studies comparing CVD risk assessment models that included ECG findings with those that did not include ECG findings were also eligible. Studies were required to report reclassification (ability to correctly reassign persons into clinically meaningful risk categories), calibration (agreement between observed and predicted outcomes), or discrimination (ability to distinguish between persons who will vs. will not have an event). Detailed descriptions of the specific test performance measures are provided in Table 1. These measures assess performance of risk prediction models or the comparative performance of models. Studies that only assessed the association between ECG findings and outcomes (e.g., with adjusted hazard ratios) were excluded. The review focused on the benefits and harms of adding ECG to the current standard practice of CVD risk prediction using traditional risk factors: male sex, older age, cigarette smoking, hypertension, dyslipidemia (high total cholesterol, high low-density lipoprotein cholesterol, or low HDL-C), and diabetes. In current clinical practice, the PCE or FRS is typically used for risk prediction. Studies were not required to specifically use the PCE or FRS to be eligible, although such studies have greatest applicability to current practice. Eligible base models included age, sex, systolic blood pressure, antihypertensive medication use, total cholesterol, HDL, and current smoking or restricted samples to remove some of these variables. Models were not required to include diabetes or race/ethnicity, but models were eligible that

Screening for CVD Ri sk Wi th ECG 7 RTI–UNC EPC included them. Comparisons that would not allow us to isolate the effect attributable to ECG were not eligible.

For KQ 3 (harms), prospective cohort studies, large retrospective cohort studies, and well- designed case-control studies (only for rare events) were also eligible. Eligible harms included mortality, , CV events, or injuries from exercise ECG; ; labeling; and harms of subsequent procedures or interventions initiated as a result of screening (e.g., subsequent angiography or revascularization procedures resulting in harm). For harms of subsequent procedures/interventions, studies that compare the procedure/intervention with no procedure/intervention were also eligible. For studies reporting rates of harms from exercise ECG or subsequent procedures/interventions, large registries or multicenter studies without a control group that report rates of harms for asymptomatic persons were also eligible. Two investigators independently reviewed titles and abstracts; those marked for potential inclusion by either reviewer were retrieved for evaluation of the full text. Two investigators independently reviewed the full text to determine final inclusion or exclusion. Disagreements were resolved by discussion and consensus.

Quality Assessment and Data Abstraction

The quality of trials and cohort studies was assessed as good, fair, or poor, using predefined criteria developed by the USPSTF and adapted for this topic (Appendix B3). For risk prediction studies (KQ 2), predefined criteria from the Checklist for Critical Appraisal and Data Extraction for Systematic Review of Prediction Modelling Studies (CHARMS)43 were used and adapted for this topic (Appendix B3). Two independent investigators assigned quality ratings for each study. Disagreements were resolved by discussion. Only studies rated as having good or fair quality were included.

For each included study, one investigator extracted pertinent information about the methods, populations, interventions, comparators, outcomes, timing, settings, and study designs. A second team member reviewed all data extractions for completeness and accuracy.

Data Synthesis and Analysis

Findings for each KQ were qualitatively synthesized by summarizing the characteristics and results of included studies in tables, figures, and narrative format. To determine whether meta- analyses were appropriate, the clinical and methodological heterogeneity of the studies was assessed following established guidance.44 The populations, tests, treatments, comparators, outcomes, and study designs were assessed qualitatively, looking for similarities and differences. At least three similar studies had to be available to estimate pooled effects. For KQ 1, pooled effects were not estimated because fewer than three similar studies were found, but risk ratios and 95 percent confidence intervals were calculated for binary outcomes reported by the included RCTs and a forest plot showing the results was created. For KQ 2, considerable heterogeneity of ECG findings assessed, base prediction models used, outcomes (e.g., all-cause mortality, CV mortality, CVD events, fatal ischemic heart disease), and duration of followup was

Screening for CVD Ri sk Wi th ECG 8 RTI–UNC EPC found; therefore, the results are presented in tabular format and in figures. Results were stratified by ECG findings evaluated, separating results for exercise ECG and resting ECG. Within the studies of resting ECG results were stratified to separate those that evaluated the addition of a constellation of ECG abnormalities from those that evaluated single/specific ECG changes. Results were categorized by the base models used as “published coefficient models,” meaning the model preserved the coefficients of original published models that have been externally validated (e.g., FRS or PCE), or as “model development.” For KQ 2, the C-statistic (Harrell’s C) and AUC were used as the primary measures of discrimination and were summarized together. Measures of overall performance were summarized with those of calibration. Net reclassification improvement (NRI) was the primary measure of reclassification, with event and nonevent NRIs reported separately when possible. There is no guidance in the literature about how to qualitatively characterize the magnitude or clinical meaning of changes in discrimination. To describe the magnitude of changes in discrimination, the following definitions were employed for practical reasons. For changes in the C-statistic, 0.1 or greater was considered large, 0.05-0.1 was considered to be moderate, 0.025-0.05 was considered small, and changes less than 0.025 were considered very small. C-statistics range from 0.5 to 1.0; the 0.1 cutpoint for large was set because it represents 20 percent of the possible range. A change in C-statistic of 0.025 approximates a 5 percent higher sensitivity when specificity is 50 percent. Analyses were conducted and figures were produced using Stata version 14 (StataCorp) and Microsoft Excel.

Two independent reviewers assessed the overall strength of the body of evidence for each KQ as high, moderate, low, or insufficient using methods developed for the USPSTF (based on methods of the EPC program45, 46), based on the overall quality of studies, consistency of results between studies, precision of findings, and risk of reporting bias. The applicability of the findings to U.S. primary care populations and settings was also assessed. Discrepancies were resolved through consensus discussion.

Expert Review and Public Comment

A draft report was reviewed by content experts, representatives of federal partners, USPSTF members, and AHRQ Medical Officers and was revised based on comments, as appropriate. It was also posted for public comment. Based on the comments received we revised the report to add additional summary sections within the results; description and interpretation of measures for discrimination, calibration, and reclassification; description of prior USPSTF recommendations; and clarification that this review was limited to studies assessing resting or exercise ECG, along with discussion of the findings of another meta-analysis that evaluated any screening test for in persons with .

USPSTF Involvement

This review was funded by AHRQ. AHRQ staff and members of the USPSTF participated in developing the scope of work and reviewed draft reports, but the authors are solely responsible for the content.

Screening for CVD Ri sk Wi th ECG 9 RTI–UNC EPC Chapter 3. Results

Literature Search

In total, 4,595 unique records were identified and 524 full texts assessed for eligibility (Figure 2). In total, 507 articles were excluded for various reasons detailed in Appendix C, and 16 studies (described in 17 articles) of good or fair quality were included. Of the included studies, two were studies of the benefits of ECG screening (KQ 1); 14 were studies of reclassification, calibration, or discriminat ion; and one study was of harms of ECG screening (KQ 3, which was also included for KQ 1). Compared with the previous evidence review for the USPSTF,42 the current review includes three studies47-49 that were in both reviews and 13 studies that are only in the current review. The previous review included many additional studies reporting associations (e.g., adjusted hazard ratios) between ECG findings and outcomes that were not eligible for this review because they did not report discrimination, calibration, or reclassification. It also included two studies related to harms of exercise ECG that were not eligible for the current review (one was an uncontrolled single center report of military officers getting stress tests;50 the other described survey data on harms, focusing on symptomatic participants51); both studies are described in the discussion of this report. Details of quality assessments of included studies and studies excluded because of poor quality are provided in Appendix D.

Results

KQ 1a. Does the Addition of Screening With Resting or Exercise ECG Improve Health Outcomes Compared With Traditional CVD Risk Factor Assessment Alone in Asymptomatic Adults? 1b. Does Improvement in Health Outcomes Vary for Subgroups Defined by Baseline CVD Risk, Age, Sex, or Race/Ethnicity?

Summary

No eligible studies compared screening with resting ECG and no screening, and none evaluated the use of screening with ECG for the purpose of risk reclassification to inform decisions about preventive medications. Two RCTs (DYNAMIT and DADDY-D) with a total of 1,151 participants evaluated screening with exercise ECG in high risk, asymptomatic adults ages 50 to 75 years with diabetes. Both RCTs reported no statistically significant improvement in health outcomes, but were limited by not reaching sample size targets.

Detailed Results: Characteristics of Included Trials

We included two fair-quality RCTs: Do You Need to Assess Myocardial Ischemia in Type-2 diabetes (DYNAMIT)52 and Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients (DADDY-D).53 The characteristics of the included studies are summarized in Table 2. Both trials compared screening with exercise ECG with no screening exercise ECG.

Screening for CVD Ri sk Wi th ECG 10 RTI–UNC EPC The DYNAMIT study was a multicenter randomized trial that randomized 631 ambulatory patients who consulted a diabetes specialist to screening versus no screening.52 Those in the screening arm were referred for detection of silent ischemia using a bicycle exercise test or Dipyridamole Single Photon Emission Computed Tomography (SPECT). SPECT was used in patients unable to perform the exercise test, with a submaximal negative exercise test, or with ECG abnormalities impairing the interpretation of the exercise test (31% ultimately had SPECT). Those with positive exercise or SPECT tests were referred to cardiologists, and all subsequent investigations and treatments were left to the judgment of the cardiologists (i.e., no protocol for that part of the process related to angiography vs. no angiography; pragmatic approach). The DADDY-D trial randomized 520 participants from a single center (2 diabetes outpatient clinics) to screening or no screening.53 Participants were required to have a normal ECG to get into the study. Those in the screening arm underwent maximal symptom-limited exercise treadmill test (ETT). Submaximal tests were considered not diagnostic and did not lead to any further investigations. Coronary angiography was proposed to all patients with positive ETTs, and choices to perform stenting or surgery were reportedly determined according to the European Guidelines by two interventional cardiologists and a cardiac surgeon after reviewing coronary anatomy.

DYNAMIT was conducted in France, and the DADDY-D trial was conducted in Italy. Mean duration of followup for both trials was 3.5 to 3.6 years. Both enrolled ambulatory patients with a clinical diagnosis of type 2 diabetes. Mean hemoglobin A1cs were 8.6 (DYNAMIT) and 7.7 (DADDY-D), respectively. DYNAMIT enrolled patients ages 55 to 75 years; mean age was 64. DADDY-D enrolled patients ages 50 to 70 years; mean age was 62. Less than half of the participants (20–45%) were women in both trials. Neither trial reported information about race or ethnicity of participants. DADDY-D did not report baseline prevalence of hypertension (but 74% were on antihypertensive medications) or PAD; they were 89 percent and 14 percent for DYNAMIT, respectively. The prevalence of heart failure in both trials was less than 1 percent. Less than half of the participants (17–39%) in both trials were smokers.

Both trials were rated as fair quality. Neither trial reached the sample size targets. DYNAMIT was stopped early because of trouble recruiting and a lower than expected event rate (it randomized 631 of the planned 3,000). DADDY-D aimed for 364 per group but enrolled about 260 per group; because the target number of participants could not be achieved, the followup period was extended from two years to 3.5 years for those who had been enrolled (the authors reported a power of 77%). For DADDY-D, masking of outcome assessors was not reported and amount of attrition was unclear.

Results of Included Trials

The main results are shown in Figure 3 and Appendix E Table 1. Overall, neither study found a statistically significant reduction in any category of events for screening compared with no screening, including their primary composite outcomes, all-cause mortality, CV-related mortality, MI, heart failure, or stroke, although findings were imprecise.

In the DYNAMIT trial, 28 participants in the screening group and 26 in the unscreened arm experienced at least one primary endpoint (composite of death from all causes, nonfatal MI,

Screening for CVD Ri sk Wi th ECG 11 RTI–UNC EPC nonfatal stroke, or heart failure requiring hospitalization or emergency service intervention) (hazard ratio [HR], 1.00; 95% confidence interval [CI], 0.59 to 1.71). In the same trial, 13 in the screening arm and 15 in the unscreened arm experienced coronary events (fatal or nonfatal MI, hospitalized , or heart failure requiring hospitalization or emergency service intervention) (HR, 0.77; 95% CI, 0.37 to 1.63). In the DADDY-D trial, 12 in the screened arm and 14 in the unscreened arm experienced a primary outcome, cardiac events defined as a composite of nonfatal MI or cardiac death (HR, 0.85; 95% CI, 0.39 to 1.84).

Subgroup analyses were performed by gender, age, and CV risk for multiple outcomes in the DADDY-D trial. No statistically significant differences were found between groups based on gender, age, or CV risk for the primary outcome of cardiac events. For heart failure, no significant differences were found between those screened and those not screened based on gender, age, or having a CV risk >20, but a difference was found for those with a CV risk <20 (0 vs. 5 events, p=0.022). For cardiac death, no significant differences were found between those screened and those not screened based on gender, CV risk, or being less than 60 years of age, but a difference was found for those >60 years of age. Fewer cardiac deaths were reported for those older than 60 years who were screened than those who were not screened (0 vs. 4 events, p=0.044). No significant differences were found between those screened and not screened based on gender, age, or CV risk for nonfatal MI. The study did not report interaction tests for the subgroup analyses, no adjustments were made for multiple testing, and it is unclear whether the subgroup analyses were planned a priori. The positive subgroup analysis findings are likely due to chance.

KQ 2. Does the Addition of Screening With Resting or Exercise ECG to Traditional CVD Risk Factor Assessment Accurately Reclassify Persons Into Different Risk Groups or Improve Measures of Calibration and Discrimination?

Summary

Fourteen good- or fair-quality studies were included.47 Five evaluated exercise ECG47, 48, 54-56 and nine evaluated resting ECG.49, 57-64 Of those nine, five evaluated multiple ECG changes (a group of major and minor changes)49, 57-60 and four evaluated only single ECG changes.61-64 Of the studies evaluating exercise ECG, three used published coefficient base models (2 FRS and 1 SCORE).47, 48, 55 Of the studies evaluating resting ECG, five reported some analyses using published coefficient base models (5 FRS; 1 also used PCE).49, 57-59, 63

For exercise ECG, although evidence from five cohort studies (9,582 participants) shows that the addition of exercise ECG abnormalities to traditional CVD risk factors results in small improvements in discrimination (absolute improvement in AUC or C-statistics 0.02 to 0.03), it is uncertain whether calibration or appropriate risk classification improves. For resting ECG, evidence from nine cohort studies (66,407 participants) shows that the addition of ECG findings to traditional CVD risk factors results in very small or small improvements in discrimination (absolute improvement in AUC or C-statistics 0.001 to 0.05) and improvements for calibration and appropriate risk classification for prediction of multiple outcomes (e.g., all-cause mortality,

Screening for CVD Ri sk Wi th ECG 12 RTI–UNC EPC CVD mortality, CHD events, or CVD events), but evidence was limited by imprecision, quality, considerable heterogeneity, and the risk categories evaluated.

Characteristics of Studies Evaluating Exercise ECG

Five cohort studies evaluated whether adding exercise ECG to traditional CVD risk prediction could improve discrimination, calibration, or risk reclassification (Tables 3 and 4).47, 48, 54-56 The studies evaluated a total of 9,582 participants. Sample sizes ranged from 98854 to 3,554.47 One enrolled participants in the 1970s;56 the other four enrolled participants in the 1990s. Two studies were conducted in the United States,47, 54 two in France,48, 55 and one in .56 Mean duration of followup ranged from 6 to 8 years in four studies; one had 26 years of followup.56 The five studies used data from five different cohorts (although the 2 studies from France may have some overlap in a subset of the participants). For the base model, two used the FRS (with published coefficients of the original model),48, 55 one used the European SCORE (with published coefficients),47 one model development study used FRS variables,54 and one model development study used some of the traditional risk factors (age, total cholesterol, systolic blood pressure, and smoking status) and restricted its sample to eliminate other risk factors (excluded women, those with prevalent diabetes, and those on blood pressure-lowering therapy at baseline) but did not account for HDL.56

All were prospective cohorts with participants from cardiology or prevention centers in urban hospitals. Participants were self-referred or referred by providers to a preventive cardiology unit in two studies,48, 55 participants presented for executive physicals in one study,47 participants were a subset of those from a study evaluating coronary artery calcium and SPECT in one study,54 and participants were recruited from five governmental agencies in one study.56 Four of the studies reported that all participants were asymptomatic. One study reported that 16.5 percent of participants had atypical symptoms and that the participants were a subset of persons having both coronary artery calcium score (CACS) and SPECT for “clinically indicated reasons”;54 it is unclear what proportion of participants were truly asymptomatic and what the clinically indicated reasons for testing were. The mean age of participants ranged from 50 to 58 years. Most participants in all trials were men, with the proportion of female participants ranging from 0 to 38 percent. Four studies did not report information about race/ethnicity; one reported that 2 percent of participants were nonwhite.47 The baseline prevalence of hypertension and diabetes ranged from 0 to 55 percent and 0 to 11 percent, respectively. The percentage of smokers ranged from 10 to 47 percent. Mean baseline FRS score was 10.8 to 12.3 in studies reporting it.48, 54, 55

Study end points included all-cause mortality,47 coronary events (cardiac deaths, sudden deaths, acute MI, and stable or unstable angina),48, 55 cardiac events (cardiac death, nonfatal MI, and the need for coronary revascularization following the development of symptomatic CAD),54 and CHD mortality (deaths caused by ischemic heart disease and sudden, unexpected deaths).56

All five of the included studies received fair-quality ratings (Appendix D). The most common methodological concerns were not reporting CIs for calibration or discrimination (5/5), not reporting measures of reclassification (4/5), selective inclusion of participants in the model based on data availability (4/5), unknown masking of outcome assessors (4/5), unknown if predictors

Screening for CVD Ri sk Wi th ECG 13 RTI–UNC EPC were assessed masked for the outcome (4/5), not reporting both discrimination and calibration (3/5), unclear handling and amount of missing data (2/5), uncertain validity and reliability of method used for measuring outcomes (2/5), using base model equations that have not been externally validated (i.e., model development studies) (2/5), and mean duration of followup less than 10 years (4/5), despite the risk prediction being focused on 10 years. The study with the longest followup (26 years) did not account for one of the traditional risk factors (HDL).56 In addition, the only study reporting reclassification did not use the risk categories commonly used in current practice for making treatment decisions (it used <6% vs. 6 to 20% vs. >20%) and may have included many symptomatic participants (16.5% had atypical chest pain and participants were a subset of those having CACS and SPECT for “clinically indicated reasons”).54

Results of Studies Evaluating Exercise ECG

For the comparison of interest to this review, one model development study reported reclassification,54 four reported calibration,48, 54-56 and three reported discrimination.47, 54, 55 Results of the included studies are shown in Table 5 and Figures 4–7. The frequency of abnormal exercise tests across included studies ranged from 6.4 to 16.7 percent (Appendix E Table 3).

Discrimination

Three studies reported discrimination for the addition of exercise ECG variables to traditional risk factors;47, 54, 55 one of the three used FRS with published coefficients for the base model55 and two were model development studies. Main results are shown in Figures 4 and 5, illustrating the AUC or C-statistic for the base model (black squares in the figures) and the AUC or C- statistic for the base model plus exercise ECG (white squares in the figures). Figure 4 is limited to the studies that used FRS or PCE with published coefficients for the base model whereas Figure 5 also shows model development studies. In addition to the AUC or C-statistic results, the columns of the figures show the outcome, ECG findings evaluated, base model used, sample size, and number of participants with an event. All three studies reported small absolute improvements in AUC or C-statistics (0.02 to 0.03), and none of them reported CIs for the discrimination data. One of the three reported a p-value indicating no statistically significant difference between a model with exercise testing variables and the base model with FRS variables (p=0.3).54

Calibration or Overall Performance

Four studies reported calibration or overall performance of models that added exercise ECG results to traditional risk factors (Table 5).48, 54-56 Two of the studies used the FRS (with published coefficients) in base models48, 55 and two were model development studies.54, 56 None of the studies reported figures such as calibration plots, but one provided a table of predicted and observed events for quintiles of risk.56 All four studies reported different measures: likelihood ratio test;48 Akaike information criteria (AIC), Brier’s score, and Hosmer-Lemeshow χ2;55 global χ2;54 and numbers of predicted and observed events.56

The two studies that used the FRS were both conducted in France and focused on prediction of

Screening for CVD Ri sk Wi th ECG 14 RTI–UNC EPC coronary events.48, 55 One study (1,051 participants) reported that model performance was not improved for the full sample with the addition of symptom-limited exercise ECG to the model (p=0.13).48 For the subgroup with pretest Framingham risk of 10.4 percent or greater, the authors reported a statistically significant improvement for adding exercise ECG to some base models, but not when the base model was FRS (n=526, p=0.06). The other study (2,709 participants) reported improved goodness of fit with the addition of exercise ECG, indicated by lower AIC (748.9 vs. 727.8) and Brier’s score (0.035 vs. 0.033); the Hosmer-Lemeshow test showed no difference between the models (p, 0.99 vs. 0.99).55

Two studies developed new models rather than using published coefficients of existing models (e.g., FRS and PCE). The study described below under Risk Reclassification (Chang et al, 201554) reported better calibration for the model that included exercise testing results than for the base model (global chi-square 16.16 vs. 11.72, p=0.04).54 The study conducted in Norway (2,014 participants) reported numbers of predicted and observed events for models with and without exercise testing (Table 5).56 Although both models (with and without exercise testing) show steep and similar gradients in age-adjusted CHD mortality (indicating good calibration), the model with exercise testing showed a slightly steeper gradient across quintiles of risk, with slightly better correspondence of predicted and observed events.

Risk Reclassification

One study (Chang et al, 201554; 988 participants) reported on the reclassification from adding exercise ECG to traditional CVD risk factor assessment for predicting cardiac events.54 It did not use the current common clinical thresholds to reclassify risk; instead, it used categories defined by 10-year risk of <6 percent, 6 to 20 percent, and >20 percent. For the base model, the authors were not able to calculate the FRS as published because blood pressure and cholesterol measurements were not available (so these predictors were dichotomized based on history of and hypertension). Therefore, it was considered a model development study with FRS variables in the base model (i.e., rather than a study using the externally validated FRS coefficients in the base model). The study found that adding the presence or absence of stress- induced ischemia detected during symptom-limited exercise treadmill testing to the base model increased the total NRI in subjects overall (9.6%; p=0.007) and in the intermediate risk group (18.9%; p=0.01). It did not report event NRI and nonevent NRI. The study also reported absolute and relative integrated discrimination improvement (IDI). For all patients, the IDI was small but statistically significant (absolute IDI, 1.4%, p=0.006; relative IDI, 110%, p<0.0001). The IDI was also improved significantly for the intermediate risk category (absolute IDI%, 1.7 [p=0.01]; relative IDI, 92% [p=0.004]). The study authors also reported on calibration, finding better calibration for the model that included exercise testing results than for the base model (global chi-square 16.16 vs. 11.72, p=0.04). However, adding exercise testing variables to the base model did not significantly improve discrimination (change in AUC, 0.02, p=0.3).

Characteristics of Studies Evaluating Resting ECG

Nine studies of resting ECG met inclusion criteria (Tables 6 and 7).49, 57-64 Eight were cohort studies and one used data from an RCT.49 Five evaluated multiple ECG changes: either a constellation of major and minor ECG changes based on the Minnesota code or an ECG risk

Screening for CVD Ri sk Wi th ECG 15 RTI–UNC EPC equation (that included multiple ECG changes).49, 57-60 Four only evaluated single ECG changes.61-64 Some studies that evaluated multiple ECG changes also reported findings for single/specific ECG changes. Three studies reported results from the National Health and Nutrition Examination Survey (NHANES),57, 59, 63 two studies reported results from the Atherosclerosis Risk in Community study,62, 64 and the remaining studies reported results from the Health, Aging, Body Composition Study,58 the Women’s Health Initiative,49 the Jichi Medical School Cohort,61 and the Copenhagen City Heart Study.60 Excluding double-counted populations, the studies evaluated a total of 68,475 subjects. Sample sizes ranged from 1,26449 to 15,375.62 One study enrolled participants in the 1970s;60 seven enrolled participants in the late 1980s and/or the 1990s, and one used a derivation cohort from the 1970s and a validation cohort from the 1980s and 1990s.59 Seven studies were conducted in the United States, one in Japan,61 and one in Denmark.60 Duration of followup ranged from 649 to 19 years.59

For the base model, three studies used the FRS (with published coefficients of the original model),49, 57, 63 one used the FRS and the PCE (with published coefficients; it also included model development analyses using FRS variables),59 one focused on model development (with traditional risk factors) but also used FRS (with published coefficients) in some secondary analyses,58 and four were model development studies using FRS variables (one of those also included intake and heart rate61).60-62, 64

All were population-based studies. Overall, the studies provided little or no information about any evaluation of whether participants had symptoms at baseline. It is unclear what proportion of participants were truly asymptomatic. One study reported excluding those with angina or claudication.57 Another study included participants with symptoms of angina or claudication, counting them among the 5 percent with prevalent CHD enrolled in the study.62 The studies either excluded those with a history of CVD or enrolled a small percentage of persons with a history of CVD. The mean age of participants ranged from 54 to 73 years. The majority of participants in all studies were women and one study enrolled only women.49 Three studies did not report information about race/ethnicity; the range of nonwhite participants in those that did report race/ethnicity was 9 to 41 percent. The baseline prevalence of diabetes ranged from 0 to 13 percent. The percentage of smokers ranged from 22 to 54 percent. Mean baseline CVD risk was not reported by most studies.

Study end points included all-cause mortality,57, 59, 60 CV mortality,57, 59, 60, 63 CHD events,49, 58, 60, 64 stroke events,61 CVD events,49, 60 sudden cardiac death,62 and ischemic heart disease mortality. 59

One study was rated as good quality,58 and the others were rated as fair (Appendix D). The most common methodological concerns were selective inclusion of participants in the model based on data availability (9/9), unknown masking of outcome assessors (8/9), unknown if predictors were assessed masked for the outcome (7/9), not reporting CIs for either calibration or discrimination (5/9), not reporting calibration (5/9), using base model equations that have not been externally validated (i.e., model development studies) (4/9),58, 60, 61, 64 not reporting the amount of missing data (2/9), and mean duration of followup less than 10 years (2/9).49, 58 In addition, of the studies reporting reclassification, just three included a threshold between risk categories corresponding to the current USPSTF recommendations for initiating preventive medications (i.e., 7.5% or 10%

Screening for CVD Ri sk Wi th ECG 16 RTI–UNC EPC 10-year risk).57-59 Of the three, one reported using cut points between categories of 1, 5, and 10 percent;59 one used <5, 5 to 9.9, 10 to 19.9, and 20 or greater;57 and one used <7.5 percent, 7.5 percent to 15 percent, and >15 percent.58 The study that used cut points of 1, 5 and 10 percent cited the European Society of Cardiology recommendations (from 2012) for lipid management as the rationale for the chosen cut points; those recommendations are based on SCORE (which predicts CVD death), and the recommendations describe that a 5 percent SCORE risk of CVD death equates to a 10 to 25 percent 10-year FRS risk of total CVD, depending on which of the several Framingham models is chosen.65 One study based NRI cut points for risk categories on the distribution of the data rather than using a priori cut points.60 One study included alcohol use and heart rate in addition to traditional risk factors in its base model.61

Among the five studies that evaluated multiple ECG changes, the most common methodological concerns (Appendix D) were selective inclusion of participants in the model based on data availability (5/5), unknown masking of outcome assessors (4/5), unknown if predictors were assessed masked for the outcome (4/5), not reporting CIs for either calibration or discrimination (3/5), not reporting the amount of missing data (2/5), not reporting calibration (2/5), using base model equations that have not been externally validated (i.e., model development studies) (2/5),58, 60 and mean duration of followup less than 10 years (2/5).49, 58 In addition, two studies reporting reclassification did not use the risk categories commonly used in current practice in the United States for making treatment decisions.59, 60 One study based NRI cut points for risk categories based on the distribution of the data rather than using a priori cut points.60

Results of Studies Evaluating Resting ECG: Multiple ECG Changes

Five studies evaluated the effect of adding major or minor ECG abnormalities or an ECG risk equation (that included multiple ECG abnormalities) to traditional risk factors.49, 57-60 Four of the five evaluated the incremental improvement of adding major or minor ECG abnormalities to the FRS or PCE (with published coefficients) for some of the outcomes reported.49, 57-59 Three studies evaluated CHD events,49, 58, 60 two studies evaluated CVD events,49, 60 three studies evaluated CVD mortality,57, 59, 60 three studies evaluated all-cause mortality,57, 59, 60 and one evaluated ischemic heart disease mortality.59 The frequency of ECG abnormalities across these studies ranged from 31 to 55 percent (Appendix E Table 4).

Discrimination

All five studies reported discrimination for the addition of ECG variables to traditional risk factors. Three of the five reported it using FRS or PCE with published coefficients.49, 57, 59 Main results are shown in Figure 4 and Figure 5, illustrating the AUC or C-statistic for the base model (black squares in the figures) and the AUC or C-statistic for the base model plus ECG changes (white squares in the figures). Figure 4 is limited to the studies that used FRS or PCE with published coefficients for the base model whereas Figure 5 also shows model development studies. In addition to the AUC or C-statistic results, the columns of the figures show the outcome, ECG findings evaluated, base model used, sample size, and number of participants with an event. All five studies reported very small57 to small absolute improvements in AUC or C-statistics. Two studies reported p-values, with one study indicating statistically significant differences between models with ECG abnormalities and a base model with conventional risk

Screening for CVD Ri sk Wi th ECG 17 RTI–UNC EPC factors for outcomes of fatal CVD and combined fatal and nonfatal CVD (p<0.001);60 the other study approached statistical significance (p=0.05).57

Calibration or Overall Performance

Three studies reported calibration or overall performance of models that added multiple changes from resting ECG to traditional risk factors (Appendix E Table 2).49, 57, 58 Two of the studies used the FRS (with published coefficients) in base models,49, 57 and one was a model development study (it reported reclassification for the addition of ECG variables to FRS, but only reported calibration for a model with ECG and FRS variables).58 None of the studies reported figures such as calibration plots. The measures reported included likelihood ratio χ2 test,49, 58 Hosmer-Lemeshow χ2,57, 58 and Bayes information criterion (BIC).57

The two studies that used the FRS were both conducted in the United States and focused on prediction of CV mortality57 or CHD events and CVD events.49 One study (6,025 participants) using NHANES data reported that the addition of major and minor ECG changes improved calibration and performance for predicting CV mortality (Hosmer-Lemeshow χ2 decreased from 15.14 to 10.98, p-values 0.05 and 0.2, respectively; BIC decreased from 3,360.54 to 3,358.28).57 The other study (1,264 participants) reported that the addition of major and minor ECG abnormalities to FRS improved model performance for predicting both CHD and CVD events (likelihood ratio chi square test: p=0.004 and p=0.02, respectively).

The one model development study that reported calibration for a model with ECG and FRS variables was conducted in the United States and used the Health ABC Study cohort of 2,192 adults ages 70 to 79 years.58 The study reported that a model with traditional risk factors (FRS variables) did not show a good calibration and was not improved by the addition of ECG abnormalities (Hosmer-Lemeshow χ2 increased with the addition of ECG abnormalities from 67.6 to 87.9, likelihood ratio p<0.00005, goodness of fit p-values 0.03 for FRS variables vs. 0.01 for FRS variables plus ECG).58

Risk Reclassification

Four of the five studies evaluating multiple ECG changes reported results on reclassification when adding resting ECG to traditional risk factors (Appendix E Table 2; Figures 6 and 7).57-60 Figures 6 and 7 show the NRI results, with black squares indicating the total NRI (sum of the event NRI and nonevent NRI), gray squares indicating the event NRI (net upward reclassification among persons who had an event), and white squares representing nonevent NRI (net downward reclassification among persons who did not have an event). For some studies, only the total NRI (black square) is provided because the data for event and nonevent NRI were not reported. Figure 6 shows model development studies and studies that used FRS or PCE with published coefficients for the base model, whereas Figure 7 is limited to studies that used FRS or PCE with published coefficients for the base model. In addition to the NRI results, the columns of the figures show the outcome, ECG findings evaluated, base model used, sample size, number of participants with an event, and the risk categories used for analyses. Three of the studies used the FRS (with published coefficients) in base models,57-59 one also used PCE (with published coefficients) and a model with FRS variables (not using published coefficients),59 and one was a

Screening for CVD Ri sk Wi th ECG 18 RTI–UNC EPC model development study.60 All four studies reported NRI and all but one60 provided event NRI or nonevent NRI data (or the table to allow us to calculate it) for some models. Two studies reported IDI (Appendix E Table 2).58, 59

For studies using published coefficient models, Figure 7 shows the net reclassification results. Three studies reported outcomes of reclassification and IDI. One study of adults in the NHANES population used four risk categories (<5%, 5 to 10%, 10 to 20%, and >20%) to calculate reclassification indices.57 Investigators reported an overall NRI of 3.6 percent, where the NRI for those with events was 3 percent and the NRI for those without events was 0.6 percent, indicating a net appropriate reclassification of those with CVD mortality upward into higher risk categories. This study also reported a clinical NRI for intermediate risk patients of 13.6 percent, although this estimate was not corrected for bias. 66 The absolute value of the IDI was low, although statistically significant. (0.0001, p<0.001). A study of older adults in the Healthy Aging and Body Composition (Healthy ABC) study also reported an improved overall NRI and IDI when ECG abnormalities were added to FRS (NRI 5.7% [95% CI, -0.4 to 11.6], IDI, 1.03% [95% CI, 0.56% to 1.50%]), although specific NRIs for events and nonevents were not provided for the comparison with FRS.58 The third study evaluated the addition of an ECG risk equation (that included frontal T axis, QT interval, and heart rate) to FRS, PCE, or a model with traditional risk factors.59 It used NHANES I data to derive the risk equation and validated the model with NHANES III.59 The clinical risk thresholds were based on the European Society of Cardiology (1%, 5%, and 10%). Adding the ECG risk equation to the FRS, PCE, or conventional risk factors resulted in improved classification. Categorical NRIs ranged from 4 to 30 percent, and continuous NRI ranged from 33 to 57 percent. Absolute IDI ranged from 0.2 to 2.6 percent, and relative IDIs ranged from 7 percent to 47 percent.

Three model development studies examined improvement in CV risk prediction when any ECG abnormality was added to conventional CV risk factors, which included diabetes in addition to the risk factors of the FRS variables.58-60 Two of them also evaluated published coefficient models (FRS or PCE) as described in the previous paragraph.58, 59 Two studies were conducted in older adult populations with mean age ranging from 70 to 74 years. The Healthy ABC study looked at the outcome of CHD,58 while a study of the Copenhagen City Heart Study examined outcomes of fatal CVD events, fatal and nonfatal CVD events, and all-cause mortality.60 The third study used NHANES cohorts to evaluate the addition of an ECG risk score to several base models.59 All three studies reported improvements in discrimination and reclassification for all outcomes examined.

The models using the Healthy ABC cohort reported an NRI for those with events of -0.9 percent, indicating inappropriate reclassification of patients with CHD events to lower risk categories. The NRI for persons without events was 8.3 percent, indicating appropriate reclassification of patients without CHD events to lower risk categories. The study authors reported on reclassification of intermediate risk participants using an unadjusted and adjusted clinical NRI (13.6% and 6.7%, respectively). The Copenhagen City Heart Study reported improved categorical NRIs and continuous NRIs, although event and nonevent component NRIs were not reported for the former (Appendix E Table 2 and Figure 6).60 The study that evaluated the addition of an ECG risk equation (that included frontal T axis, QT interval, heart rate, age, and sex) to a model with traditional risk factors used NHANES I data to derive the risk equation and

Screening for CVD Ri sk Wi th ECG 19 RTI–UNC EPC validated the model with NHANES III.59 The clinical risk thresholds were based on the European Society of Cardiology (1%, 5%, and 10%). Adding the ECG risk equation to conventional risk factors resulted in improved classification. Categorical NRIs ranged from 4 percent (for fatal ischemic heart disease) to 11 percent (for all-cause mortality), with reclassification to higher categories for persons with events accounting for most of the total NRI (i.e., event NRIs were greater than nonevent NRIs).

Results of Studies Evaluating Resting ECG: Single ECG Changes

Five studies evaluated the effect of adding a single ECG abnormality to traditional risk factors. Studies evaluated a range of outcomes including CVD events,60 CHD events,64 stroke,61 CVD mortality,60, 63 sudden cardiac death,62 and all-cause mortality.60 Three studies evaluated changes and other ventricular abnormalities, three studies examined LVH, and the remaining two studies looked at a variety of ECG changes. The frequency of ECG abnormalities across the studies that evaluated single ECG changes ranged from 1 to 24 percent (Appendix E Table 4).

T Wave Changes and Other Ventricular Repolarization Abnormalities

One study using NHANES III data examined the effect of adding observed T wave amplitude greater than -0.2 mV in lead aVR to the FRS on risk prediction of CV mortality. 63 Investigators observed an improvement in discrimination (C-statistic 0.812 to 0.820), good calibration, and overall categorical NRI of 0.07 using clinical risk categories of <5 percent, 5 to 10 percent, 10 to 20 percent, and >20 percent. Net reclassification of subjects with events was 2.7 percent, and net reclassification of subjects without events was 2.3 percent, indicating participants were appropriately reclassified to higher or lower risk categories. IDI was reported as 0.012 (p<0.01).

The Copenhagen City Heart study examined the addition of any T wave changes or ventricular conduction delay (VCD) to conventional CV risk factors. For both repolarization abnormalities, the C-statistic improved from 0.705 to 0.716 (T wave) or 0.706 (VCD) for the outcome of fatal CVD events. For the combined outcome of nonfatal and fatal CVD events, discrimination improved from 0.651 to 0.658 for any T wave abnormality, but there was no change for VCD. Calibration results were not reported. Different risk thresholds were used for outcomes of fatal CVD events and fatal and nonfatal CVD events (Appendix E Table 2 and Figure 6) when calculating categorical NRIs, and separate event and nonevent NRIs were not reported. Adding T wave changes to conventional risk factors resulted in an overall continuous NRI of 29.2 percent and a categorical NRI of 5.4 percent for fatal CVD events and 20.3 percent and 2.7 percent for fatal and nonfatal CVD events, respectively. Addition of VCD to conventional risk factors resulted in small but significant NRIs for fatal CVD events (overall continuous NRI of 12.1% and categorical NRI of 1.1%) but nonsignificant or no change for fatal and nonfatal CVD events.

Finally, prolonged QTc interval was added to conventional risk factors to participants in the Jichi Medical School Cohort. Discrimination and calibration results were not reported. Adding prolonged QTc to conventional risk factors resulted in a categorical NRI of 0.026 (event NRI, 1.35%; nonevent NRI, 1.22%) and nonsignificant IDI (0.291; p=0.80).61

Screening for CVD Ri sk Wi th ECG 20 RTI–UNC EPC Left Ventricular Hypertrophy

Three studies added LVH to conventional risk factors.60, 61, 64 In the Copenhagen City Heart study, the addition of LVH to conventional risk factors resulted in increased discrimination of fatal CVD events and continuous and categorical NRIs of 2.8 percent and 1.1 percent, respectively.60 There was no improvement in discrimination or significant NRI findings for fatal or nonfatal CVD events. Calibration and IDI results were not reported for this study. In a study of the Atherosclerosis Risk in Communities (ARIC) cohort, LVH was examined as a continuous measure using the Cornell score, as well as a categorical variable, by gender and diabetes subgroups.64 For women, adding continuous or categorical LVH ECG findings to conventional risk factors worsened discrimination of CHD events (0.707 and 0.709, respectively vs 0.711) among those with diabetes and did not change discrimination in those without diabetes (0.777 for all models). For men, there was no or minimal improvement in discrimination. Calibration, reclassification, and IDI outcomes were not reported. The Jichi Medical School Cohort did not report discrimination or calibration outcomes but did report a categorical NRI of 0.020 (event NRI, 1.01%; nonevent NRI, 1.01%) and IDI of 0.004 (p=0.75) for a model that included LVH.61

Other ECG Changes

The remaining studies evaluated different individual ECG changes combined with conventional risk factors.60, 62 One study examined the effect of adding the finding of a deep terminal negativity of the P wave in lead V1 to conventional risk factors and reported a categorical NRI of 0.028 for clinical risk thresholds of 5 percent and 15 percent.62 Event NRI was 0.028, indicating participants who had events were primarily reclassified to higher risk categories. Discrimination, calibration, and IDI were not reported.

In the Copenhagen City Heart Study, Q waves, ST segment depressions, and resting heart rate were added to conventional risk factors with improvements in discrimination compared with conventional risk factors alone for fatal CVD events and fatal or nonfatal CVD events60 (Appendix E Table 2). However, categorical NRIs were only significant for ST segment depressions (3.1% for fatal CVD, 2.2% for all CVD events), and event and nonevent NRIs were not reported separately. Calibration and IDI were also not reported.

KQ 3a. What Are the Harms of Screening With Resting or Exercise ECG, Including Harms of Subsequent Procedures or Interventions Initiated as a Result of Screening? 3b. Do the Harms of Screening Vary for Subgroups Defined by Baseline CVD Risk, Age, Sex, or Race/Ethnicity?

Summary

One eligible study reported on harms from subsequent procedures or interventions initiated as a result of screening. It reported that one person out of 12 undergoing revascularization had a nonfatal MI. No other eligible studies reported rates of harms from screening asymptomatic adults with resting or exercise ECG. We searched for, but did not find, other studies evaluating

Screening for CVD Ri sk Wi th ECG 21 RTI–UNC EPC potential harms such as mortality, arrhythmia, CV events, injuries, anxiety, labeling, and harms of subsequent procedures or interventions initiated as a result of screening.

Detailed Results

One of the fair-quality RCTs described in KQ 1, the DADDY-D trial, reported some results eligible for this KQ. 53 Twenty out of 262 participants (7.6%) in the screened group had positive ETTs. Of those 20, 17 underwent coronary angiography (6.5% of the 262 in the screened group). Angiography revealed critical stenosis (not defined) in 71 percent (12/17), and all patients with critical stenosis underwent revascularization procedures (7 percutaneous and 5 surgical). One patient having percutaneous revascularization had a nonfatal acute MI 3 days after the procedure and underwent a second percutaneous angioplasty. His ejection fraction was reported to be normal 6 months after the event.

The other trial described in KQ 1 (DYNAMIT) reported the number of some subsequent tests but did not report whether any of the tests or interventions resulted in harms; adverse events that occurred during followup were not recorded.52 Sixty-eight of the 316 participants (21.5%) in the screened group had a definitely abnormal or an uncertain screening test (exercise test or SPECT) result. Of those, 38 underwent coronary angiography (12% of the 316 in the screened group) and nine subsequently underwent coronary angioplasty (7 of those 9 received stents) and three had coronary artery bypass graft.

Screening for CVD Ri sk Wi th ECG 22 RTI–UNC EPC Chapter 4. Discussion

Summary of Evidence

Tables 8 and 9 provide a summary of findings in this evidence review. Table 8 is organized by KQ and provides a summary of the main findings along with a description of consistency, precision, quality, limitations, strength of evidence, and applicability. For KQ 2, the summary of evidence for exercise ECG and resting ECG was separated. The overall strength of evidence was low or insufficient for each of the questions and outcomes evaluated. No RCTs of screening with resting ECG were found; RCTs of exercise ECG in asymptomatic participants found no improvement in health outcomes despite focusing on higher risk populations with diabetes, although they were limited by not reaching sample size targets. Scant direct evidence on harms of screening asymptomatic persons with ECG was found. Evidence on whether the addition of exercise ECG to traditional CVD risk factors results in accurate reclassification is lacking. Cohort studies found that the addition of multiple resting ECG abnormalities to traditional CVD risk factors accurately reclassifies persons, and improves discrimination and calibration, but evidence was limited by imprecision, quality, considerable heterogeneity, and inconsistent use of risk thresholds that align with clinical decisions and recommendations.

Evidence for the Benefits and Harms of Screening With Resting or Exercise ECG

No eligible studies compared screening with resting ECG and no screening, and none evaluated the use of screening with ECG for the purpose of risk reclassification to inform decisions about whether to initiate preventive medications. Two RCTs (DYNAMIT and DADDY-D, total of 1,151 participants) evaluated screening with exercise ECG in asymptomatic adults ages 50 to 75 years with diabetes compared with no screening and found no statistically significant improvement in health outcomes, including their primary composite outcomes, all-cause mortality, CV-related mortality, MI, heart failure, or stroke. Findings from the two studies were consistent but imprecise. For the primary composite outcomes, HRs were 1.00 (95% CI, 0.59 to 1.71) for a composite of death from all causes, nonfatal MI, nonfatal stroke, or heart failure requiring hospitalization or emergency service intervention and 0.85 (95% CI, 0.39 to 1.84) for a composite of nonfatal MI or cardiac death. Some key limitations of the trials include not reaching sample size targets and stopping early because of trouble recruiting. Both trials followed participants for about 3.5 years, and longer followup may be needed to adequately evaluate screening with exercise ECG. The overall strength of evidence for whether screening with exercise ECG improves health outcomes was low (for no benefit) because of imprecision and risk of bias.

The participants in the included trials were higher risk groups that would be, in theory, more likely to benefit from screening with exercise ECG to identify silent ischemia. For example, in DYNAMIT, participants had diabetes plus two of the following additional risk factors: urinary albumin excretion above a threshold, hypertension, hyperlipidemia, history of PAD (14%), history of TIA (4 to 5%), tobacco consumption, or family history of premature CVD. However,

Screening for CVD Ri sk Wi th ECG 23 RTI–UNC EPC even among these higher risk groups of asymptomatic diabetics, screening with exercise ECG (followed by referral to cardiology [DYNAMIT] or recommendation for coronary angiography [DADDY-D] for those with abnormal exercise ECGs) did not improve health outcomes.

Potential harms of screening asymptomatic adults with resting or exercise ECG include mortality, arrhythmia, CV events, injuries, anxiety, labeling, and harms of subsequent procedures or interventions initiated as a result of screening (e.g., subsequent angiography or revascularization procedures resulting in harm). Both DYNAMIT and DADDY-D reported numbers of subsequent tests and interventions after abnormal exercise tests, but just one (DADDY-D) reported whether any of the tests or interventions resulted in harms (1/12 [8.3%] undergoing percutaneous revascularization had a nonfatal acute MI 3 days later). Among asymptomatic diabetics, DADDY-D and DYNAMIT reported that 7.6 percent (20/262 screened) had abnormal ETTs and 21.5 percent (68/316 screened) had definitely abnormal or uncertain bicycle exercise or SPECT results, respectively. Most of those with abnormal screening tests underwent coronary angiography (DADDY-D: 17/262, 6.5%; DYNAMIT 38/316, 12%), and some had revascularization procedures (12/262, 4.6% and 12/316, 3.8%, respectively).

No other eligible studies reported rates of harms for asymptomatic adults. Studies without control groups were eligible if they were multicenter studies or registries that reported rates of harms from exercise ECG or subsequent procedures/interventions specifically for asymptomatic persons. A single site study of 377 asymptomatic military officers (mean age 37) at one station in the Midwestern United States reported that none had complications during exercise testing.50 Of the 377, 45 (11.9%) had abnormal exercise test results and 10 (2.7%) underwent coronary angiography. Of those, one was reported to have “mild CAD.” Many other studies have reported rates of angiography (but no information on harms) for asymptomatic persons after exercise ECG; rates have ranged from 0.6 percent to 13 percent, although most reported rates less than 3 percent.47, 48, 50, 67-74 Rates of subsequent revascularization have also been reported by some, with those studies estimating lower rates than reported by DADDY-D and DYNAMIT. For example, two studies (with 3,554 and 1,051 participants, respectively) reported rates of 0.1 percent to 0.5 percent undergoing revascularization after screening exercise ECG.47, 48 Little is known about the harms of revascularization procedures for adults without symptoms or a prior diagnosis of CVD (Appendix A). Regardless of symptom status, some tests that follow an abnormal ECG expose patients to radiation, including coronary angiography, computed tomography angiography, and myocardial imaging.75 Coronary angiography can expose patients to as much radiation as 600 to 800 chest X-rays.76

Studies that focused on symptomatic adults have reported rates of harms of exercise ECG and harms of subsequent procedures/interventions. The scientific statement from the AHA with recommendations for clinical exercise laboratories reports that the rate is usually considered to be approximately 1 in 10,00051 (0.01%). It references a review of eight studies estimating sudden cardiac death during exercise testing that reported rates from zero to 0.005 percent (5 per 100,000 tests).51, 77 The statement also notes that survey data provide estimates of rates for complications: hospitalization including serious arrhythmias (0.2% or less), acute MI (0.04%), or sudden cardiac death during or immediately after an exercise test (0.01%).51, 78

Screening for CVD Ri sk Wi th ECG 24 RTI–UNC EPC Reclassification, Calibration, and Discrimination With the Addition of Resting or Exercise ECG to Traditional CVD Risk Factor Assessment

No consensus exists for the thresholds that should be considered clinically significant changes in discriminat ion (e.g., AUC, C-statistic), calibration, or reclassification (e.g., NRI). Changes in these measures are important to evaluate in the context of each other, but appropriate reclassification (NRI) has the most direct clinical meaning. Nevertheless, studies must use clinically meaningful risk categories (i.e., that correspond to clinical decisions, such as 7.5% or 10% 10-year risk) to allow for the potential clinical significance of NRI results to be assessed. Further, interpreting NRI is not simple because it is unfamiliar to many and it is calculated from four proportions. For more clear interpretation, focusing on the event NRI and nonevent NRI components may help. For example, nonevent NRI is the net downward reclassification (i.e., appropriate reclassification) among persons who did not have an event; it is calculated as the proportion of persons without an event who were appropriately reclassified into a lower risk group minus the proportion of those without an event who were inappropriately reclassified into a higher risk group.

For exercise ECG, although evidence from five cohort studies (9,582 participants) shows that the addition of exercise ECG results to traditional CVD risk factors results in small improvements in discriminat ion (absolute improvement in AUC or C-statistics 0.02 to 0.03), it is uncertain whether calibration or appropriate risk classification improves. Evidence was limited by imprecision and risk of bias for all outcomes and by inconsistency or unknown consistency for calibration and reclassification outcomes. Some important limitations of the evidence include that CIs for calibration or discrimination were not reported; mean duration of followup was less than 10 years in four of the five studies; reclassification was only reported by one study; unknown masking of outcome assessors in four studies; and not reporting both discrimination and calibration in three studies. The only study reporting reclassification was a model development study (i.e., used FRS variables but did not use published coefficients) that used risk categories of <6 percent, 6 to 20 percent, and >20 percent and may have included many symptomatic participants (because 16.5% had atypical chest pain and participants were a subset of those having CACS and SPECT for “clinically indicated reasons”).54 Also, we found an absence of evidence related to exercise ECG for healthy, low risk persons (e.g., mean age was 50 to 58 and mean baseline FRS score was 10.8 to 12.3 in studies reporting it).

For resting ECG, evidence from nine cohort studies (66,407 participants) shows that the addition of ECG findings to traditional CVD risk factors results in very small or small improvements in discriminat ion (absolute improvement in AUC or C-statistics 0.001 to 0.05) and in improvements for calibration and appropriate risk classification for prediction of all-cause mortality, CVD mortality, CHD events, or CVD events. Total NRIs (event; nonevent) range from 3.6 percent (2.7%; 0.6%) to 30 percent (17%; 19%) for studies using FRS or PCE base models (95% CIs were rarely reported). However, evidence was limited by imprecision and risk of bias for all outcomes and by incomplete and inconsistent reporting of discrimination, calibration, and reclassification measures. For reclassification, although evidence consistently showed improved NRI, the estimates of NRI and the outcomes assessed were inconsistent, and the consistency of findings is unknown for specific risk thresholds because all studies used different risk categories. The body of evidence also had considerable heterogeneity in baseline prediction models used,

Screening for CVD Ri sk Wi th ECG 25 RTI–UNC EPC type of ECG abnormalities added to base models, and outcomes assessed (e.g., all-cause mortality, CVD mortality, CHD events). The reported discrimination of base models varied widely, ranging from inadequate to excellent (AUC or C-statistics from 0.58 to 0.85) likely because of the different outcomes, different patient populations, and different base models used. Only one study used the base model for risk prediction (i.e., the PCE) that the USPSTF and ACC/AHA currently recommend to inform clinical decisions about preventive medications.

An important limitation of the evidence was a lack of reporting on any assessment of symptoms; it is unclear what proportion of participants was truly asymptomatic in most of the studies of resting ECG. Perhaps the proportion with symptoms is likely to be relatively low given that the studies were population based and most of them excluded persons with a history of CVD, but it is uncertain whether enrolling even a small percentage of symptomatic participants could artificially inflate estimates of appropriate reclassification. Other important limitat ions included unknown masking of outcome assessors in eight studies, CIs for calibration or discrimination were not reported in five studies, calibration or overall performance was not reported in five studies, and not using established risk prediction models with published coefficients in four studies (i.e., model development studies).

For NRI, event and nonevent NRI components were often not reported as recommended. Although an overall positive value of NRI indicates net appropriate reclassification into appropriate risk strata, the clinical implications can be very different if the majority of patients are those with events being shifted into higher risk categories (event NRI) as opposed to those without events being shifted into lower risk categories (nonevent NRI). Although the addition of ECG abnormalities to conventional risk factors improves total NRI in both cases, one might lead to an increase in preventive medications, while the other suggests a possible reduction in the use of preventive medications.

For reclassification with resting ECG, few studies included a threshold between risk categories corresponding to the USPSTF recommendations for preventive medications (i.e., 7.5% or 10% 10-year risk). The potential for appropriate reclassification based on the addition of major and minor ECG changes to existing models (PCE or FRS) initially looks promising when viewing Figure 7 because studies reported increases in total appropriate reclassification (total NRI), appropriate reclassification of persons with events to higher risk categories (event NRI), and appropriate reclassification of persons without events to lower risk categories (nonevent NRI). However, several cautions should be noted: (1) no two studies evaluated the same model, risk category thresholds, and outcome. Therefore, none of the models that included ECG variables have been validated in more than one study; (2) no CIs were provided for most of that data; (3) NRI is highly dependent on risk category thresholds, which varied widely across these studies; (4) evaluating risk reclassification using four categories to determine NRI is potentially not clinically meaningful (or is less clinically meaningful than using fewer categories) and may artificially make the reclassification look better because each reclassification counts as a positive move for the NRI if someone with an event moves from any lower to any higher category regardless of whether the change would correspond to different treatment decisions (likewise for someone without an event who moves into any lower risk category); (5) a single study59 accounts for six of the nine rows in Figure 7. It reported NRI for three different base models for prediction of several mortality outcomes but did not evaluate prediction of CHD or CVD events

Screening for CVD Ri sk Wi th ECG 26 RTI–UNC EPC because it used NHANES data that do not have that capability. The study is limited by using risk category thresholds of 1, 5, and 10 percent and not reporting the full reclassification table to allow determination of how much of the NRI was accounted for by reclassification that should change clinical decisions (e.g., from 5 to 9.9% to 10% or greater risk for persons with events) versus how much was accounted for by reclassification that would have no effect on clinical decisions and outcomes (e.g., from 1 to 4.9% to <1% for persons without events). The study59 is also the only study that evaluated adding an ECG risk equation to base models; (6) another study58 in Figure 7 is limited by only having 7.5 years of followup. It also focused on elderly participants ages 70 to 79 years and did not report event NRI or nonevent NRI (or the data to calculate those) for the addition of ECG changes to the FRS base model (those were reported for a model development part of the study and showed net inappropriate reclassification for persons who had events, event NRI -0.9%). It is uncertain whether risk reclassification could provide clinically useful information for this population given recent evidence on lack of benefit of statins for primary prevention in elderly persons of similar age 79 and given the USPSTF I statement on initiation of aspirin for primary prevention in adults age 70 years or older.

Limitations

This review did not evaluate the evidence on preventive medications (i.e., aspirin and lipid lowering therapy) that could be initiated based on risk reclassification or the evidence on benefits and harms of lifestyle counseling to reduce CV risk. Other systematic reviews for the USPSTF have evaluated that evidence. But one of the contextual questions (Appendix A) summarizes what medications (i.e., aspirin, lipid-lowering therapy) are recommended for persons in various CVD risk categories. This review did not systematically review the benefits and harms of revascularization procedures; contextual question 2 (Appendix A) summarizes information on the harms and benefits of revascularization procedures for adults without symptoms or a prior diagnosis of CVD.

For KQ 2 (reclassification, calibration, and discrimination), it was sometimes challenging to determine whether studies used published coefficients (e.g., used the FRS or PCE) or whether they were model development studies that used the FRS variables. Study authors were contacted to clarify the uncertainty, but some did not respond. Therefore, some studies could be misclassified (regarding model development vs. published coefficients). For KQ 2, we used qualitative terms (e.g., very small, small, moderate) to describe the magnitude of changes in discriminat ion. These qualitative terms are meant only to be descriptive of specific numeric ranges and are not intended to indicate a corresponding clinically meaningful benefit for health outcomes. No consensus exists for the thresholds that should be considered clinically meaningful changes in the AUC/C-statistic or NRI.

For KQ 3 (harms of screening), for studies without control groups to be eligible for this review, studies were required to be multicenter studies or from large registries. This approach excluded a single center study of 377 asymptomatic military officers50; that study is described above in the Discussion. This review did not evaluate harms from ECG or subsequent procedures/interventions for symptomatic populations.

Screening for CVD Ri sk Wi th ECG 27 RTI–UNC EPC This review was limited to studies assessing resting ECG or exercise ECG. A previously published meta-analysis evaluated any screening test for coronary artery disease in persons with type 2 diabetes.80 It identified five trials (including DYNAMIT and DADDY-D) with a total of 3,315 participants. The trials that were not eligible for our review evaluated stress scintigraphy, coronary CT angiography, or stress echocardiography with exercise ECG. Pooled analyses found no statistically significant association with all-cause mortality (RR, 0.95; 95% CI, 0.66 to 1.35) or cardiac events (RR, 0.72; 95% CI, 0.49 to 1.06).

Future Research Needs

To better understand whether risk classification is improved in a clinically useful way that is likely to improve health outcomes, risk prediction studies that evaluate the addition of ECG abnormalities to the PCE (as the base model) would be most useful because the PCE is the approach currently recommended by the USPSTF and ACC/AHA to assess 10-year risk and to inform decisions about preventive medications. Only one included study used the PCE as the base model. Studies of a constellation of resting ECG changes (e.g., major and minor changes based on the Minnesota code) show greater promise than those of single ECG changes and should likely be the focus of future research. Future risk studies should use clinically meaningful risk categories that correspond to recommendations about preventive medications to determine how many persons are appropriately reclassified in a manner that would lead to additional or fewer preventive medication treatments. Specifically, when considering the USPSTF recommendations for statins and aspirin, evaluating NRI related to the 10% 10-year risk threshold is of great interest. Future studies should evaluate asymptomatic populations (with some assessment of symptom status to avoid enrolling those with angina, atypical chest pain, or dyspnea, for example) and should exclude those with a history of CVD. Measures of discrimination, calibration, and reclassification (including total NRI, event NRI, and nonevent NRI) and their corresponding CIs should be reported.

Conclusion

The overall strength of evidence was low or insufficient for each of the questions and outcomes evaluated. Controlled trials of screening with exercise ECG in asymptomatic diabetic patients found no improvement in health outcomes over about 3.5 years but were limited by not reaching sample size targets and stopping early because of trouble recruiting. No controlled trials evaluated screening with resting ECG for asymptomatic adults. Potential harms of exercise ECG include arrhythmias, acute MI, and sudden cardiac death. Potential harms of both exercise and resting ECG include harms of subsequent angiography or revascularization procedures after an abnormal test, but little evidence exists to determine their frequency in asymptomatic persons. Some evidence suggests that the addition of exercise ECG to traditional risk factors results in small improvements in discrimination, but it is uncertain whether calibration or appropriate risk classification improves. Cohort studies found that the addition of multiple resting ECG findings to traditional risk factors improves discriminat ion (small absolute improvement in AUC or C- statistics), calibration, and appropriate risk classification, but evidence was limited by

Screening for CVD Ri sk Wi th ECG 28 RTI–UNC EPC imprecision, quality, considerable heterogeneity, and unknown consistency for specific risk thresholds because studies used different risk categories.

Screening for CVD Ri sk Wi th ECG 29 RTI–UNC EPC References

1. Moyer VA. Screening for coronary heart disease with electrocardiography: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 2012 Oct 2;157(7):512-8. doi: 10.7326/0003-4819-157-7-201210020-00514. PMID: 22847227. 2. U. S. Preventive Services Task Force, Bibbins-Domingo K, Grossman DC, et al. Statin use for the primary prevention of cardiovascular disease in adults: US Preventive Services Task Force Recommendation Statement. JAMA. 2016 Nov 15;316(19):1997- 2007. doi: 10.1001/jama.2016.15450. PMID: 27838723. 3. World Health Organization. Global atlas on cardiovascular disease prevention and control. Geneva: World Health Organization; 2011. 4. World Health Organization. International Statistical Classfication of and Related Health Problems, 10th Revision (ICD-10). Geneva: World Health Organization; 1992. 5. Abbate R, Cioni G, Ricci I, et al. Thrombosis and acute coronary syndrome. Thromb Res. 2012 Mar;129(3):235-40. doi: 10.1016/j.thromres.2011.12.026. PMID: 22281070. 6. Moreira DM, da Silva RL, Vieira JL, et al. Role of vascular in coronary artery disease: potential of anti-inflammatory drugs in the prevention of atherothrombosis. Inflammation and anti-inflammatory drugs in coronary artery disease. Am J Cardiovasc Drugs. 2015 Feb;15(1):1-11. doi: 10.1007/s40256-014-0094-z. PMID: 25369900. 7. Bentzon JF, Otsuka F, Virmani R, et al. Mechanisms of plaque formation and rupture. Circ Res. 2014 6;114(12):1852-66. doi: 10.1161/CIRCRESAHA.114.302721. PMID: 24902970. 8. Santos-Gallego CG, Picatoste B, Badimon JJ. Pathophysiology of acute coronary syndrome. Curr Atheroscler Rep. 2014 Apr;16(4):401. doi: 10.1007/s11883-014-0401-9. PMID: 24504549. 9. Stone GW, Maehara A, Lansky AJ, et al. A prospective natural-history study of coronary atherosclerosis. N Engl J Med. 2011 Jan 20;364(3):226-35. doi: 10.1056/NEJMoa1002358. PMID: 21247313. 10. Saab F, Mukherjee D, Gurm H, et al. Risk Factors in first presentation acute coronary syndromes (ACS): how do we move from population to individualized risk prediction? . 2009 Dec-2010 Jan;60(6):663-7. doi: 10.1177/0003319709333870. PMID: 19729368. 11. D'Agostino RB, Sr., Grundy S, Sullivan LM, et al. Validation of the Framingham coronary heart disease prediction scores: results of a multiple ethnic groups investigation. JAMA. 2001 Jul 11;286(2):180-7. PMID: 11448281. 12. Expert Panel on Detection Evaluation, Treatment of High Blood Cholesterol in Adults. Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA. 2001 May 16;285(19):2486-97. PMID: 11368702. 13. Mozaffarian D, Benjamin EJ, Go AS, et al. Heart disease and stroke statistics--2015 update: a report from the American Heart Association. Circulation. 2015 Jan 27;131(4):e29-322. doi: 10.1161/CIR.0000000000000152. PMID: 25520374.

Screening for CVD Ri sk Wi th ECG 30 RTI–UNC EPC 14. Vasan RS, Sullivan LM, Wilson PW, et al. Relative importance of borderline and elevated levels of coronary heart disease risk factors. Ann Intern Med. 2005 Mar 15;142(6):393-402. PMID: 15767617. 15. Helfand M, Buckley DI, Freeman M, et al. Emerging risk factors for coronary heart disease: a summary of systematic reviews conducted for the U.S. Preventive Services Task Force. Ann Intern Med. 2009 Oct 06;151(7):496-507. PMID: 19805772. 16. American Heart Association, American College of Cardiology. Pooled Cohort Equations Cardiovascular Risk Calculator. 2014. http://tools.acc.org/ASCVD-R isk-Estimator/. Accessed May 19, 2017. 17. Bibbins-Domingo K, U. S. Preventive Services Task Force. Aspirin use for the primary prevention of cardiovascular disease and colorectal cancer: U.S. Preventive Services Task Force Recommendation Statement. Ann Intern Med. 2016 Jun 21;164(12):836-45. doi: 10.7326/M16-0577. PMID: 27064677. 18. Ford ES, Will JC, Mercado CI, et al. Trends in predicted risk for atherosclerotic cardiovascular disease using the pooled cohort risk equations among US adults from 1999 to 2012. JAMA Intern Med. 2015 Feb;175(2):299-302. doi: 10.1001/jamainternmed.2014.6403. PMID: 25485596. 19. Dehmer SP, Maciosek MV, Flottemesch TJ. Aspirin Use to Prevent Cardiovascular Disease and Cancer: A Decision Analysis. AHRQ Publication 15-05229-EF-1. Rockville, MD: Agency for Healthcare Research and Quality; 2015. 20. Benjamin EJ, Blaha MJ, Chiuve SE, et al. Heart disease and stroke statistics-2017 update: a report from the American Heart Association. Circulation. 2017 Mar 07;135(10):e146- e603. doi: 10.1161/CIR.0000000000000485. PMID: 28122885. 21. Fang J, Shaw KM, Keenan NL. Prevalence of coronary heart disease--United States, 2006-2010. MMWR Morb Mortal Wkly Rep. 2011 Oct 14;60(40):1377-81. PMID: 21993341. 22. Fang J, Shaw KM, George M. Prevalence of stroke--United States, 2006-2010. MMWR Morb Mortal Wkly Rep. 2012 May 25;61(20):379-82. PMID: 22622094. 23. Lowres N, Neubeck L, Redfern J, et al. Screening to identify unknown atrial . A systematic review. Thromb Haemost. 2013 Aug;110(2):213-22. doi: 10.1160/th13-02- 0165. PMID: 23595785. 24. Cholesterol Treatment Trialists Collaboration, Baigent C, Blackwell L, et al. Efficacy and safety of more intensive lowering of LDL cholesterol: a meta-analysis of data from 170,000 participants in 26 randomised trials. Lancet. 2010 Nov 13;376(9753):1670-81. doi: 10.1016/S0140-6736(10)61350-5. PMID: 21067804. 25. Heart Protection Study Collaborative Group. MRC/BHF Heart Protection Study of cholesterol lowering with in 20,536 high-risk individuals: a randomised placebo-controlled trial. Lancet. 2002 Jul 06;360(9326):7-22. doi: 10.1016/S0140- 6736(02)09327-3. PMID: 12114036. 26. LaRosa JC, Grundy SM, Waters DD, et al. Intensive lipid lowering with in patients with stable coronary disease. N Engl J Med. 2005 Apr 07;352(14):1425-35. doi: 10.1056/NEJMoa050461. PMID: 15755765. 27. National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of

Screening for CVD Ri sk Wi th ECG 31 RTI–UNC EPC High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002 Dec 17;106(25):3143-421. PMID: 12485966. 28. Robinson JG, Wang S, Smith BJ, et al. Meta-analysis of the relationship between non- high-density lipoprotein cholesterol reduction and coronary heart disease risk. J Am Coll Cardiol. 2009 Jan 27;53(4):316-22. doi: 10.1016/j.jacc.2008.10.024. PMID: 19161879. 29. Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014 Jun 24;129(25 Suppl 2):S1-45. doi: 10.1161/01.cir.0000437738.63853.7a. PMID: 24222016. 30. Chou R, High Value Care Task Force of the American College of Physicians. Cardiac screening with electrocardiography, stress echocardiography, or myocardial perfusion imaging: advice for high-value care from the American College of Physicians. Ann Intern Med. 2015 Mar 17;162(6):438-47. doi: 10.7326/M14-1225. PMID: 25775317. 31. American Academy of Family Physicians. Annual EKGs for Low-risk Patients. 2012. http://www.aafp.org/patient-care/clinical-recommendations/all/cw-ekg.html. Accessed 18 Nov, 2015. 32. Lim LS, Haq N, Mahmood S, et al. Atherosclerotic cardiovascular disease screening in adults: American College of Preventive Medicine position statement on preventive practice. Am J Prev Med. 2011 Mar;40(3):381 e1-10. doi: 10.1016/j.amepre.2010.11.021. PMID: 21335273. 33. Anderson KM, Odell PM, Wilson PW, et al. Cardiovascular disease risk profiles. Am Heart J. 1991 Jan;121(1 Pt 2):293-8. PMID: 1985385. 34. Anderson KM, Wilson PW, Odell PM, et al. An updated coronary risk profile. A statement for health professionals. Circulation. 1991 Jan;83(1):356-62. PMID: 1984895. 35. American Academy of Family Physicians. Family physician interpretation of electrocardiograms. Leawood, KS: American Academy of Family Physicians; 2007. 36. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. 6th ed. Philadelphia: Lippincott Williams & Wilkins; 2000. 37. Wilson PW, D'Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation. 1998 May 12;97(18):1837-47. PMID: 9603539. 38. D'Agostino RB, Sr., Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the . Circulation. 2008 Feb 12;117(6):743- 53. doi: 10.1161/CIRCULATIONAHA.107.699579. PMID: 18212285. 39. National Heart, Lung, and Blood Institute. 2013 Report on the Assessment of Cardiovascular Risk: Full Work Group Report Supplement: National Heart, Lung, and Blood Institute; 2013. 40. Shillinglaw B, Viera AJ, Edwards T, et al. Use of global coronary heart disease risk assessment in practice: a cross-sectional survey of a sample of U.S. physicians. BMC Health Serv Res. 2012 Jan 24;12:20. doi: 10.1186/1472-6963-12-20. PMID: 22273080. 41. Bhatia RS, Bouck Z, Ivers NM, et al. Electrocardiograms in Low-Risk Patients Undergoing An Annual Health Examination. JAMA Intern Med. 2017 Jul 10doi: 10.1001/jamainternmed.2017.2649. PMID: 28692719. 42. Chou R, Arora B, Dana T, et al. Screening asymptomatic adults with resting or exercise electrocardiography: a review of the evidence for the U.S. Preventive Services Task

Screening for CVD Ri sk Wi th ECG 32 RTI–UNC EPC Force. Ann Intern Med. 2011 Sep 20;155(6):375-85. doi: 10.7326/0003-4819-155-6- 201109200-00006. PMID: 21930855. 43. Moons KG, de Groot JA, Bouwmeester W, et al. Critical appraisal and data extraction for systematic reviews of prediction modelling studies: the CHARMS checklist. PLoS Med. 2014 Oct;11(10):e1001744. doi: 10.1371/journal.pmed.1001744. PMID: 25314315. 44. West SL, Gartlehner G, Mansfield AJ, et al. Comparative Effectiveness Review Methods: Clinical Heterogeneity Methods Research Paper. AHRQ Publication No. 10-EHC070-EF. Rockville, MD: Agency for Healthcare Research and Quality; September 2010. http://effectivehealthcare.ahrq.gov/ 45. Owens DK, Lohr KN, Atkins D, et al. AHRQ series paper 5: grading the strength of a body of evidence when comparing medical interventions--Agency for Healthcare Research and Quality and the Effective Health-Care Program. J Clin Epidemiol. 2010 May;63(5):513-23. doi: 10.1016/j.jclinepi.2009.03.009. PMID: 19595577. 46. Berkman ND, Lohr KN, Ansari MT, et al. Grading the strength of a body of evidence when assessing health care interventions: an EPC update. J Clin Epidemiol. 2015 Nov;68(11):1312-24. doi: 10.1016/j.jclinepi.2014.11.023. PMID: 25721570. 47. Aktas MK, Ozduran V, Pothier CE, et al. Global risk scores and exercise testing for predicting all-cause mortality in a preventive medicine program. JAMA. 2004 Sep 22;292(12):1462-8. doi: 10.1001/jama.292.12.1462. PMID: 15383517. 48. Cournot M, Taraszkiewicz D, Galinier M, et al. Is exercise testing useful to improve the prediction of coronary events in asymptomatic subjects? Eur J Cardiovasc Prev Rehabil. 2006 Feb;13(1):37-44. PMID: 16449862. 49. Denes P, Larson JC, Lloyd-Jones DM, et al. Major and minor ECG abnormalities in asymptomatic women and risk of cardiovascular events and mortality. JAMA. 2007 Mar 7;297(9):978-85. doi: 10.1001/jama.297.9.978. PMID: 17341712. 50. Hollenberg M, Zoltick JM, Go M, et al. Comparison of a quantitative treadmill exercise score with standard electrocardiographic criteria in screening asymptomatic young men for coronary artery disease. N Engl J Med. 1985 Sep 5;313(10):600-6. doi: 10.1056/NEJM198509053131003. PMID: 4022047. 51. Myers J, Arena R, Franklin B, et al. Recommendations for clinical exercise laboratories: a scientific statement from the american heart association. Circulation. 2009 Jun 23;119(24):3144-61. doi: 10.1161/CIRCULATIONAHA.109.192520. PMID: 19487589. 52. Lievre MM, Moulin P, Thivolet C, et al. Detection of silent myocardial ischemia in asymptomatic patients with diabetes: results of a randomized trial and meta-analysis assessing the effectiveness of systematic screening. Trials. 2011;12:23. doi: 10.1186/1745-6215-12-23. PMID: 21269454. 53. Turrini F, Scarlini S, Mannucci C, et al. Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients? The DADDY-D trial. Screening diabetic patients for unknown coronary disease. Eur J Intern Med. 2015 Jul;26(6):407-13. doi: 10.1016/j.ejim.2015.05.006. PMID: 26058988. 54. Chang SM, Nabi F, Xu J, et al. Value of CACS compared with ETT and myocardial perfusion imaging for predicting long-term cardiac outcome in asymptomatic and symptomatic patients at low risk for coronary disease: clinical implications in a multimodality imaging world. JACC Cardiovasc Imaging. 2015 Feb;8(2):134-44. doi: 10.1016/j.jcmg.2014.11.008. PMID: 25677886.

Screening for CVD Ri sk Wi th ECG 33 RTI–UNC EPC 55. Cournot M, Taraszkiewicz D, Cambou JP, et al. Additional prognostic value of , exercise testing, and arterial ultrasonography for coronary risk assessment in primary prevention. Am Heart J. 2009 Nov;158(5):845-51. doi: 10.1016/j.ahj.2009.08.017. PMID: 19853707. 56. Erikssen G, Bodegard J, Bjornholt JV, et al. Exercise testing of healthy men in a new perspective: from diagnosis to prognosis. Eur Heart J. 2004 Jun;25(11):978-86. doi: 10.1016/j.ehj.2004.04.009. PMID: 15172470. 57. Badheka AO, Patel N, Tuliani TA, et al. Electrocardiographic abnormalities and reclassification of cardiovascular risk: insights from NHANES-III. Am J Med. 2013 Apr;126(4):319-26.e2. doi: 10.1016/j.amjmed.2012.10.020. PMID: 23415052. 58. Auer R, Bauer DC, Marques-Vidal P, et al. Association of major and minor ECG abnormalities with coronary heart disease events. JAMA. 2012 Apr 11;307(14):1497-505. doi: 10.1001/jama.2012.434. PMID: 22496264. 59. Shah AJ, Vaccarino V, Janssens AC, et al. An electrocardiogram-based risk equation for incident cardiovascular disease from the National Health and Nutrition Examination Survey. JAMA Cardiol. 2016 Aug 3doi: 10.1001/jamacardio.2016.2173. PMID: 27487404. 60. Jorgensen PG, Jensen JS, Marott JL, et al. Electrocardiographic changes improve risk prediction in asymptomatic persons age 65 years or above without cardiovascular disease. J Am Coll Cardiol. 2014 Sep 2;64(9):898-906. doi: 10.1016/j.jacc.2014.05.050. PMID: 25169175. 61. Ishikawa J, Ishikawa S, Kario K. Prolonged corrected QT interval is predictive of future stroke events even in subjects without ECG-diagnosed left ventricular hypertrophy. Hypertension. 2015 Mar;65(3):554-60. doi: 10.1161/hypertensionaha.114.04722. PMID: 25534703. 62. Tereshchenko LG, Henrikson CA, Sotoodehnia N, et al. Electrocardiographic deep terminal negativity of the P wave in V(1) and risk of sudden cardiac death: the Atherosclerosis Risk in Communities (ARIC) study. J Am Heart Assoc. 2014 Dec;3(6):e001387. doi: 10.1161/jaha.114.001387. PMID: 25416036. 63. Badheka AO, Patel NJ, Grover PM, et al. ST-T wave abnormality in lead aVR and reclassification of cardiovascular risk (from the National Health and Nutrition Examination Survey-III). Am J Cardiol. 2013 Sep 15;112(6):805-10. doi: 10.1016/j.amjcard.2013.04.058. PMID: 23764245. 64. Folsom AR, Chambless LE, Duncan BB, et al. Prediction of coronary heart disease in middle-aged adults with diabetes. Diabetes Care. 2003 Oct;26(10):2777-84. PMID: 14514579. 65. Perk J, De Backer G, Gohlke H, et al. European Guidelines on cardiovascular disease prevention in clinical practice (version 2012). The Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J. 2012 Jul;33(13):1635-701. doi: 10.1093/eurheartj/ehs092. PMID: 22555213. 66. Paynter NP, Cook NR. A bias-corrected net reclassification improvement for clinical subgroups. Med Decis Making. 2013 Feb;33(2):154-62. doi: 10.1177/0272989X12461856. PMID: 23042826.

Screening for CVD Ri sk Wi th ECG 34 RTI–UNC EPC 67. Blumenthal RS, Becker DM, Yanek LR, et al. Detecting occult coronary disease in a high-risk asymptomatic population. Circulation. 2003 Feb 11;107(5):702-7. PMID: 12578872. 68. Boyle RM, Adlakha HL, Mary DA. Diagnostic value of the maximal ST segment/heart rate slope in asymptomatic factory populations. J Electrocardiol. 1987 Oct;20 Suppl:128- 34. PMID: 3320257. 69. Davies B, Ashton WD, Rowlands DJ, et al. Association of conventional and exertional coronary heart disease risk factors in 5,000 apparently healthy men. Clin Cardiol. 1996 Apr;19(4):303-8. PMID: 8706370. 70. Dunn RL, Matzen RN, VanderBrug-Medendorp S. Screening for the detection of coronary artery disease by using the exercise tolerance test in a preventive medicine population. Am J Prev Med. 1991 Sep-Oct;7(5):255-62. PMID: 1790029. 71. Livschitz S, Sharabi Y, Yushin J, et al. Limited clinical value of exercise stress test for the screening of coronary artery disease in young, asymptomatic adult men. Am J Cardiol. 2000 Aug 15;86(4):462-4. PMID: 10946046. 72. Massie BM, Szlachcic Y, Tubau JF, et al. Scintigraphic and electrocardiographic evidence of silent coronary artery disease in asymptomatic hypertension: a case-control study. J Am Coll Cardiol. 1993 Nov 15;22(6):1598-606. PMID: 8227826. 73. Piepgrass SR, Uhl GS, Hickman JR, Jr., et al. Limitations of the exercise stress test in the detection of coronary artery disease in apparently healthy men. Aviat Space Environ Med. 1982 Apr;53(4):379-82. PMID: 7082255. 74. Pilote L, Pashkow F, Thomas JD, et al. Clinical yield and cost of exercise treadmill testing to screen for coronary artery disease in asymptomatic adults. Am J Cardiol. 1998 Jan 15;81(2):219-24. PMID: 9591907. 75. Fazel R, Krumholz HM, Wang Y, et al. Exposure to low-dose from medical imaging procedures. N Engl J Med. 2009 Aug 27;361(9):849-57. doi: 10.1056/NEJMoa0901249. PMID: 19710483. 76. Choosing Wisely. EKGs and exercise stress tests When you need them-and when you don’t. Philadelphia, PA: Choosing Wisely® is an initiative of the ABIM Foundation©. All rights reserved; 2017. http://www.choosingwisely.org/patient-resources/ekgs-and- exercise-stress-tests/. Accessed November 13, 2017. 77. Gordon NF, Kohl HW. Exercise testing and sudden cardiac death. J Cardiopulm Rehabil Prev. 1993;13:381-6. 78. American College of Sports Medicine. ACSM’s Guidelines for Exercise Testing and Prescription. 7th ed. Philadelphia: Lippincott Williams & Wilkins; 2006. 79. Han BH, Sutin D, Williamson JD, et al. Effect of statin treatment vs usual care on primary cardiovascular prevention among older adults: The ALLHAT-LLT Randomized Clinical Trial. JAMA Intern Med. 2017 May 22doi: 10.1001/jamainternmed.2017.1442. PMID: 28531241. 80. Rados DV, Pinto LC, Leitao CB, et al. Screening for coronary artery disease in patients with type 2 diabetes: a meta-analysis and trial sequential analysis. BMJ Open. 2017 May 09;7(5):e015089. doi: 10.1136/bmjopen-2016-015089. PMID: 28490559.

Screening for CVD Ri sk Wi th ECG 35 RTI–UNC EPC Figure 1. Analytic Framework

Figure 1. A n al yt i c Fr a mewo r k 1

Screening with resting or exercise ECG Preventive Measures of medications or lifestyle Adults without reclassification (new Health outcomes counseling† symptoms or a CVD risk after All-cause mortality 2 diagnosis of screening test), Cardiovascular mortality CVD* calibration, or Cardiovascular events discrimination 3

Harms Harms†

* Includes adults regardless of their CVD risk (those with low, intermediate, or high risk are eligible) as assessed by traditional risk factors (those included in Framingham risk models): male sex, older age, cigarette smoking, hypertension, dyslipidemia (high total cholesterol, high low-density lipoprotein cholesterol, or low high-density lipoprotein cholesterol), and diabetes. † This systematic review does not include key questions about the benefits and harms of preventive medications to reduce cardiovascular risk (i.e., aspirin and lipid-lowering therapy) or the benefits and harms of lifestyle counseling, because those have been addressed by other systematic reviews for the USPSTF.

Abbreviations: CVD = cardiovascular disease; ECG = electrocardiography; KQ – key quest ion.

Key Questions to Be Systematically Reviewed

1a. Does the addition of screening with resting or exercise electrocardiography (ECG) improve health outcomes compared with traditional cardiovascular disease (CVD) risk factor assessment alone in asymptomatic adults? 1b. Does improvement in health outcomes vary for subgroups defined by baseline CVD risk (e.g., low, intermediate, or high risk), age, sex, or race/ethnicity? 2. Does the addition of screening with resting or exercise ECG to traditional CVD risk factor assessment accurately reclassify persons into different risk groups (e.g., high-, intermediate-, and low-risk groups) or improve measures of calibration and discriminat ion? 3a. What are the harms of screening with resting or exercise ECG, including harms of subsequent procedures or interventions initiated as a result of screening? 3b. Do the harms vary for subgroups defined by baseline CVD risk (e.g., low, intermediate, or high risk), age, sex, or race/ethnicity?

Screening for CVD Ri sk Wi th ECG 36 RTI–UNC EPC Figure 2. Summary of Evidence Search and Selection Diagram

Figure 2. Summary ofE vi d enc e S c h an d Selection Diagram Number of additional unique records identified through other sources

Grey literature ((hand searching,, referencereference listlist reviewreview,, etc.):.): 6 Number of unique records identified through References from 2011 USPSTF Systematic Review of Cardiovascular database searching Disease Risk Assessment Tools:: 82 References from 2011 USPSTF Screening Asymptomatic Adults for PubMed:: 4,,036 Coronary Heart Disease With Resting or Exercise Electrocardiography:: Cochrane Library:: 287 Systematic Review to Update the 2004 U..S.. Preventive Services Task Clinicaltrials..gov and WHO ICTRP:: 121 Force Recommendation:: 63

Number of records excluded based on title Number of records screened and abstract review 4,,595 4,,071

Number of full--texttext articlesarticles excludedexcluded,, with reasons 507 Number of full--texttext articlesarticles 507 assessed for eligibility Non-English: 1 524 Non-English: 1 IneligibleIneligible populationpopulation:: 111 IneligibleIneligible//no screening//treatmenttreatment:: 112 IneligibleIneligible//no comparison:: 75 IneligibleIneligible//no outcome:: 163 IneligibleIneligible settingsetting:: 1 Number of studies ((articles)) IneligibleIneligible studystudy designdesign:: 43 includedincluded inin systematicsystematic reviewreview Poor quality:: 1 16 ((17))

2 studiesstudies 14 studiesstudies 1 studystudy includedincluded forfor includedincluded forfor includedincluded forfor KQ 1 KQ 2 KQ 3

Note:: TheThe sumsum ofof thethe numbernumber ofof studiesstudies perper KQKQ exceedsexceeds thethe totaltotal numbernumber ofof studiesstudies becausebecause somesome studiesstudies werewere applicableapplicable toto multiple KQs..

Abbreviations : ICTRP=International Clinical Trials Registry Platform; KQ=key question; USPSTF=U.S. Preventive Services Task Force; WHO=World Health Organization.

Screening for CVD Ri sk Wi th ECG 37 RTI–UNC EPC Figure 3. Main Results of Included Randomized, Controlled Trials Reporting Health Outcomes (KQ 1)

Figure 3. Ma in Resultsof IncludedRandomized , Controlled Trials Reporting Health Outcomes ( KQ 1)

DYNAMIT, primary composite outcome was defined as death from all causes, nonfatal MI, nonfatal stroke, or heart failure requiring hospitalization or emergency service intervention. DADDY-D, primary composite outcome was defined as first cardiac event, specifically nonfatal MI or cardiac death. DYNAMIT did not report data for CV-related deaths. For other CV events, the DYNAMIT trial reported no significant differences between arms for revascularization (18 vs. 21, p=0.61). The DADDY-D trial reported 19 total deaths (6 cardiac and 13 noncardiac) and seven total but did not report which group those occurred in.

Abbreviations: CV=cardiovascular; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; DADDY- D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; MI=myocardial infarction.

Screening for CVD Ri sk Wi th ECG 38 RTI–UNC EPC Figure 4. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Models

Figure 4. Effect on Discrimination of Adding Exercise ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Models

a St udy report ed c-statistic rather than AUC.

Abbreviations: AUC=area under the curve; CHD=coronary heart disease; CI=confidence interval; CVD=cardiovascular disease; ECG=electrocardiography; FRS=Framingham Risk Score; LVH=left ventricular hypertrophy; NR=not reported; PCE=pooled cohort equations.

Screening for CVD Ri sk Wi th ECG 39 RTI–UNC EPC Figure 5. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, Systemic Coronary Risk Evaluation, or Conventional Risk Factor Base Models

Figure 5. Effect on Discrimination of Adding Exercise ECG Variables to Framingham Risk Score, Pooled Cohort Equation, Systemic Coronary Risk Evaluation, or Conventional Ri sk F act o r Ba se Mo de ls

Screening for CVD Ri sk Wi th ECG 40 RTI–UNC EPC Figure 5. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, Systemic Coronary Risk Evaluation, or Conventional Risk Factor Base Models

Screening for CVD Ri sk Wi th ECG 41 RTI–UNC EPC Figure 5. Effect on Discrimination of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, Systemic Coronary Risk Evaluation, or Conventional Risk Factor Base Models

a Model also included metabolic equivalent (MET) and Duke treadmill score (DTS). b St udy report ed c-statistic rather than AUC. c In women without diabetes mellitus. d In men without diabetes mellitus.

Abbreviations: AUC=area under the curve; CHD=coronary heart disease; CI=confidence interval; CRF=conventional risk factors; CVD=cardiovascular disease; ECG=electrocardiography; FRS=Framingham Risk Score; IHD=ischemic heart disease; LVH=left ventricular hypertrophy; NR=not reported; PCE=pooled cohort equations; SCORE=Systemic Coronary Risk Evaluation.

Screening for CVD Ri sk Wi th ECG 42 RTI–UNC EPC Figure 6. Effect on Reclassification of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models

Figure 6. Effect on Reclassification of Adding Exercise ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models

Screening for CVD Ri sk Wi th ECG 43 RTI–UNC EPC Figure 6. Effect on Reclassification of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models

Screening for CVD Ri sk Wi th ECG 44 RTI–UNC EPC Figure 6. Effect on Reclassification of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models

Screening for CVD Ri sk Wi th ECG 45 RTI–UNC EPC Figure 6. Effect on Reclassification of Adding Exercise or Resting ECG Variables to Framingham Risk Score, Pooled Cohort Equation, or Conventional Risk Factor Base Models

a Model also included metabolic equivalent (MET) and Duke treadmill score (DTS).

Abbreviations: CHD=coronary heart disease; CI=confidence interval; CRF=conventional risk factors; CVD=cardiovascular disease; ECG=electrocardiography; FRS=Framingham Risk Score; LVH=left ventricular hypertrophy; NR=not reported; NRI=net reclassification index; PCE=pooled cohort equations.

Screening for CVD Ri sk Wi th ECG 46 RTI–UNC EPC Figure 7. Effect on Reclassification of Adding Resting ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Models

Figure 7. Effect on Reclassification of Adding Exercise ECG Variables to Framingham Risk Score or Pooled Cohort Equation Base Mo de ls

Abbreviations: CHD=coronary heart disease; CI=confidence interval; CVD=cardiovascular disease; ECG=electrocardiography; FRS=Framingham Risk Score; NR=not reported; NRI=net reclassification index; PCE=pooled cohort equations.

Screening for CVD Ri sk Wi th ECG 47 RTI–UNC EPC Table 1. Test Performance Measures for Comparing Risk Asse ssme nt or Prediction Models

Purpose of Outcome Example Measures Measure of Test Performance Description Discrimination c-statistic or A UC; The probability that, for a randomly selected pair of individuals, one w ith disease and the other w ithout, that the change in c-statistic or person w ith disease w ill have the higher estimated disease probability according to the model. The C-statistic can be A UC conceptualized as the area under the ROC curve (plots sensitivity against 1−specificity); as a rank order statistic, it is insensitive to systematic errors in calibration. The Harrell’s C-statistic is an extension of the AUC for survival analysis allow ing for right-censored data and variable time to follow up. The change in c-statistic or AUC can be insensitive in assessing the impact of adding new predictors to a model, and the impact of a new predictor on c-statistics is low er w hen other strong predictors are in the model. Calibration Calibration plot Graphical assessment of calibration w ith predictions on the x-axis and outcome on the y-axis. Calibration in the large and calibration slope can be derived from calibration plots. O:E The ratio of observed to expected events. Hosmer–Lemeshow χ2 Calculated by summing differences betw een observed and predicted probabilities in each group (e.g., groups defined by deciles or risk strata); a significant p-value signals poor fit. The test is sensitive to how groups are constructed and sensitive to sample size, often being nonsignificant for small N and significant for large N. Overall Akaike information Measures used during model development to aid in inclusion or exclusion of predictors in a model. The AIC is a performance criterion (AIC) and function of log likelihood that adds a penalty for each added predictor. The BIC is similar, though imposes a greater (captures both Bayes information penalty than the AIC for added variables. Low er values of both measures indicate better model fit. A change of >10 in calibration and criterion (BIC) the AIC has been proposed to indicate strong evidence for a difference in models. discrimination aspects) Likelihood ratio χ2 Likelihood ratio χ2 is a global test of model fit and is a function of the number of terms in the model. Higher values for the ratio, or difference betw een models, indicate better fit (as do low er absolute log-likelihood values). A global χ2 is generally the same as a likelihood χ2 (tw ice the log likelihood ratio). R2 There are a number of w ays to calculate an R2 for a logistic regression. Nagelkerke’s generalized R2 is generally analogous to the percentage of variance explained in a linear model and is adjusted to a range of 0 to 1. Higher values indicate better fit. The R2 is more helpful than the Brier score because it can be compared across models/studies. Brier score The Brier score computes the sum of squared differences betw een observed outcomes and fitted probability, w here low er values indicate that predicted probabilities are closer to observed outcomes. Ris k Net reclassification The sum of differences in proportions of individuals moving up (a risk category) minus those moving dow n w ith a reclassification index or improvement cardiovascular disease outcome, plus the proportion moving dow n minus those moving up w ithout an outcome. NRI (NRI) can be considered separately as the sum of the event NRI (P(up|event) - P(dow n|event)) and nonevent NRI (P(dow n|nonevent) – P(up|nonevent)). The NRI is of limited value in comparing models w ith different risk categories. Integrated Integrates the NRI over all possible cut-offs; equivalent to difference in discrimination slopes of the tw o models and to discrimination the difference in R2. improvement (IDI) Note: Table was modified with permission of the authors from the Kaiser Permanente Evidence-based Practice Center; from a table in their report on nontraditional risk factors.

Abbreviations: AIC=Akaike information criterion; AUC=area under the curve; BIC=Bayes information criterion; FRS=Framingham risk score; IDI=integrated discrimination improvement; N=sample size; NRI=net reclassification index or improvement; PCE=pooled cohort equat ions; ROC=receiver operating characteristic.

Screening for CVD Ri sk Wi th ECG 48 RTI–UNC EPC Table 2. Characteristics of Included Randomized, Controlled Trials for KQ 1 and KQ 3

% HTN % HF % TIA First Author, Years of % PAD Year G1 (N) Screening Source of Follow - Mean % Non- Mean CV Mean Mean % PV D Trial Name G2 (N) Approach Patients Country up Age (SD) % F w hite Ris k (SD) A1c (SD) BM I (SD) % Smokers Quality Lievre et al, Screened Exercise ECG, 45 hospitals, France Mean 3.5 63.9 (5.1) 45 NR NR 8.6 (2.1) 30.6 (5) 88.8 Fair 201152 (316) bicycle exercise ambulatory 0.5 DY NA MIT Not screened test (or patients w ho 4.6 (315) dipyridamole consulted a 14.1 SPECT, 31% ) a diabetes NR specialist 16.6b Turrini et al, Screened Exercise ECGc 2 diabetes Italy Mean 3.6 61.9 (5) 20 NR 20 (9)d 7.7 (2) 30.1 (6) NRe Fair 201553 (262) outpatient 0 (excluded) DA DDY -D Not screened clinics NR (258) NR 6 38.7 a SPECT was used in patients unable to perform the exercise test, with a submaximal negative exercise test, or with ECG abnormalities impairing the interpretation of the exercise test. Those with positive tests were referred to cardiologists, and all subsequent investigations and treatments were left at the cardiologist’s discretion. b T obacco consumption c Maximal symptom-limited exercise treadmill test (ETT) preformed following American Heart Association guidelines. Submaximal tests were considered not diagnost ic and did not lead to any further investigations. Coronary angiography was proposed to all patients with positive ETT; choices to perform stenting or surgery were determined according to the European Guidelines by t wo interventional cardiologists and a cardiac surgeon after reviewing coronary anatomy. d Required CV risk score ≥10% for eligibility, risk determined according to Italian risk chart (includes gender, diabetic status, age, cigarette smoking status, systolic blood pressure, serum cholesterol). e 74.3% on antihypertensive treatment; mean SBP 140.

Abbreviations: A1c=glycosylated hemoglobin; BMI=; CV=cardiovascular; DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; ECG=electrocardiogram; ETT=exercise treadmill test; F=female; HF=heart failure; HTN=hypertension; KQ=key question; G=group; N=sample size; NR=not reported; PAD=peripheral artery disease; PVD=peripheral ; SBP=systolic blood pressure; SD=standard deviation; SP ECT =Single Photon Emission Computed Tomography; T IA=transient ischemic attack.

Screening for CVD Ri sk Wi th ECG 49 RTI–UNC EPC Table 3. Characteristics of Included Studies for KQ 2 That Evaluated Exercise ECG, Part 1

Fir s t Author, Year Model Type: Countr Sam ple Years of Quality ECG Findings Evaluated Base Model Cohort Source of Patients y Size Followup Aktas, 200447 Exercise ECG according to Bruce Published From Preventive Consecutive participants US 3,554 Mean: 8 Fair (or modified Bruce) protocol; coefficient: Medicine Section of presenting for an executive ischemic ST abnormalitya using a European Cleveland Clinic physical. Self-referred. 12-lead, symptom-limited exercise SCOREb (1990–2002) ECG Chang, 201554 Ex er cise ECG according to Bruce Model Fr om the Methodist Persons w ho had both CACS US 988 Median: Fair protocol; stress-induced ischemiac development: Hospital, Houston and stress SPECT for clinically 6.9 identified via ECG during symptom- FRSd Texas (1995–2006) indicated reasons at the Heart limited exercise treadmill testing; variables and Vascular Center METS and DTS Cournot, 200648 Symptom-limited exercise ECGe Published From the preventive Consecutive asymptomatic France 1,051 Mean: 6 Fair coefficient: cardiology unit of a persons self-referred or referred FRSf teaching hospital by PCPs and cardiologists for (1995–1999) evaluation of risk factors and routine screening Cournot, 200955 A positive exercise testg during a Published From the preventive Apparently healthy France 2,709 Median: 6 Fair symptom-limited exercise ECG w ith coefficient: cardiology unit of a asymptomatic persons self- f orthogonal and V1 to V6 FRS teaching hospital referred (20%) or referred by (1996–2004) PCPs (27%) or other providers to a preventive cardiology unit Erikssen, 200456 Resting ECG and a symptom-limited Model From the University Apparently healthy males ages Norw ay Assessment 26 Fair bicycle exercise ECG testh development: Hospital of Oslo 40–60 years recruited from five 1 (1972– Classical Risk (1972–1975) governmental agencies w ho 1975): 2,014 Factor (CRF) participated in a cardiovascular Assessment regression risk assessment 2 (1980– modeli 1982): 1,428 a An ischemic ST abnormality, which was assessed visually by t wo independent readers, was defined as a 1-mm horizontal or downsloping ST-segment depression occurring 80 ms after the J-point; ST-segment depression had to be noted in at least three consecutive beats in at least t wo cont iguous leads. b SCORE includes age, sex, total cholesterol, systolic blood pressure, and smoking status (this study used the high-risk coefficients from it). c Ischemia was defined as ≥1 mm of ST-segment depression occurring >80 ms after the J-point. High and low risk were defined as the presence and absence of ischemia, respectively. d Authors attempted to calculate FRS as published, but continuous BP and cholesterol measurements were not available, so these predictors were dichotomized (hyperlipidemia defined as total cholesterol 200–239 mg/dL and HT N defined as SBP 140–159 mm Hg). e Positive ET was defined as a horizontal or downsloping ST-segment depression ≥1.0 mm at 80 ms after the J-point, in at least two contiguous leads, occurring at any time of exercise or recovery period. f Used Anderson 1991: 10-year FRS function that includes age, sex, current smoking, diabetes, total cholesterol, and HDL-C. g Positive exercise testing was defined as a horizontal or downsloping ST-segment depression ≥1.0 mm at 80 ms after the J-point, in at least two contiguous leads, occurring at any time during exercise or the recovery period. h Exercise predictors were physical fitness (cumulative work during exercise divided by body weight), maximal heart rate, systolic blood pressure at the end of the first exercise load, and exercise ECG interpretation (ST-segment depression ≥1.0 mm at 0.08 s after the J-point regardless of ST-segment morphology). i CRF model included age, total cholesterol, systolic blood pressure, and smoking status. The study included men only; therefore, sex was not needed in the model; the study also excluded persons with prevalent diabetes and persons on blood pressure–lowering therapy at baseline. High-density lipoprotein cholesterol was not accounted for in the model.

Screening for CVD Ri sk Wi th ECG 50 RTI–UNC EPC Table 3. Characteristics of Included Studies for KQ 2 That Evaluated Exercise ECG, Part 1

Abbreviations: BP=blood pressure; CACS=coronary artery calcium score; CRF=Classical Risk Factor; DTS=Duke treadmill score; ECG=electrocardiogram; ET= exercise test; FRS=Framingham Risk Score; HDL-C=high-density lipoprotein cholesterol; HTN= hypertension; METS=metabolic equivalents of task; KQ=key question; PCP=primary care physicians and cardiologists; SCORE=Systematic COronary Risk Evaluation; SPECT=single-photon emission computed tomography; U.S.=United States.

Screening for CVD Ri sk Wi th ECG 51 RTI–UNC EPC Table 4. Characteristics of Included Studies for KQ 2 That Evaluated Exercise ECG, Part 2

First Author, % With % Non- Mean BMI Year % CV D Symptoms Mean Age (SD) % F w hite CV Ris k (SD) % HTN % DM % Smokers Aktas, 200447 0 0 57 (4) 19 2 SCORE,a median (25th– 28 (4) NR (mean 3 10 75th percentile) SBP 128) 1st tertile: 0.14 (0.87–1.8) 2nd tertile: 3.0 (2.5–3.5) 3rd tertile: 6.6 (5.2–9.2) Chang, 201554 0 (for 16.5b 57.5 (9.3) 25 NR FRS, mean (SD): 11.1 NR 49.6 9.6 46.5 CA D; (6.5) NR f or Low risk (<6%): 16.9% CV D) Intermediate risk (6%– 20%): 69.2% High risk (>20%): 13.9% Cournot, 200648 0 0 Total: 51.6 (10.3) 36 NR Mean (median) FRS Total: 26.1 ≥160/95 mm 11.0 24.3 All: 12.3 (10.4) (4.5) Hg: 33.0 Negative ET, n=962: ≥140/90 mm 12.1 (10.4) Hg: 54.8 Positive ET, n=89: 14.7 (11.4) Cournot, 200955 0 0 Median: 51.6 38 NR FRS, mean (SD): 10.8 26.0 (4.4) 48.2 6.8 23.9 (10.5) (7.8) Erikssen, 200456 0 0 Assessment 1: 0 NR NR NR 0 (treated 0 Assessment 49.8 (5.5) HTN) 1: 43.8 (NR) Assessment 2: Assessment 56.6 (5.4) 2: 32.8 (NR) a SCORE provides 10-year risk for cardiovascular mortality and includes age, sex, total cholesterol, systolic blood pressure, and smoking status (this study used the high-risk coefficients from it). b Study reported 16.5% had atypical chest pain symptoms but does not report indications for other tests beyond stating that they were “clinically indicated reasons.”

Abbreviations: BMI=body mass index; CAD=coronary artery disease; CV=cardiovascular; CVD=cardiovascular disease; DM=diabetes mellitus; ECG=electrocardiogram; ET=exercise testing; F= female; FRS=Framingham Risk Score; HTN=hypertension; KQ=key quest ion; n=sample size; NR=not reported; SBP=systolic blood pressure; SCORE=Systematic COronary Risk Evaluation; SD=standard deviation.

Screening for CVD Ri sk Wi th ECG 52 RTI–UNC EPC Table 5. Results of Included Studies for KQ 2 That Evaluated Exercise ECG

Fir s t Author, Year N (%) With Quality Outcome Outcome Discrimination Calibration Reclassification Aktas, All-cause 114 (3) C-statistic (95% CI): NR NR 200447 mortality SCORE: 0.73 (NR) Fair SCORE + exercise test: 0.76 (NR), p NR Chang, Cardiac 106 (11)b A UC Global χ2 NRI % for FRS variables + ETT (p Value 201554 eventsa FRS variables: 0.63 (NR) FRS variables: 11.72 vs FRS variables) Fair FRS variables + ETT: NR FRS variables + ETTe: 16.16 All patients: 9.6 (0.007); Appropriate Use FRS variables + DTS: NR p=0.04 (FRS + ETT vs. FRS) Cohortf: 11.1 (0.005) FRS variables + METS: NR FRS variables + DTS: 14.59 Low risk: 0 (1.0); Intermediate risk: 18.9 FRS variables + ETT + METc+ p=0.24 (FRS + DTS vs. FRS) (0.01); High risk: 8.1 (0.38) DTSd: 0.65 (NR) FRS v ariables + METS: 14.68 Absolute IDI % for FRS variables + ETT p=0.3 (FRS variables + ETT + p=0.03 (FRS + METS vs. FRS) (p Value vs FRS variables) MET + DTS vs. FRS variables) All patients: 1.4 (0.006); Appropriate Use Cohort: 1.6 (0.006) Low risk: 0.1 (0.75); Intermediate risk: 1.7 (0.01); High risk: 0.88 (0.39) Relative IDI % for FRS variables + ETT (p Value vs FRS variables) All patients: 110 (<0.0001); Appropriate Use Cohort: 128 (<0.0001) Low risk: 591 (0.35); Intermediate risk: 92 (0.004); High risk: 522 (0.19) Cournot, Total 34 (3)h NR FRS Model vs. FRS Model + exercise test NR 200648 coronary (ET) results Fair eventsg ( CE) Likelihood ratio test: Whole sample (n=1051) p=0.13i Subgroup with pre-test Framingham risk ≥10.4% (n=526): p=0.06j Cournot, Definite 94 (4)l FRS vs. FRS + femoral + FRS vs. FRS + femoral bruit + positive NR 200955 coronary positive exercise testm exercise testm Fair eventsk AUROC (95% CI): 0.732 (NR) Hosmer-Lemeshow chi-square: P=0.99 vs. vs. 0.762 (NR), p NR P=0.99 Sensitivity, %: 3.2 vs. 8.5 Akaike information criterion: 748.9 vs. Specificity, %: 99.4 vs. 98.6 727.8 Positive predictive value, %: 15.8 Brier’s score: 0.035 vs. 0.033 vs. 19.1 Negative predictive value, %: 96.4 vs. 96.6

Screening for CVD Ri sk Wi th ECG 53 RTI–UNC EPC Table 5. Results of Included Studies for KQ 2 That Evaluated Exercise ECG

Fir s t Author, Year N (%) With Quality Outcome Outcome Discrimination Calibration Reclassification Erikssen, CHD 300 (15) NR CRF v s . CRF+X models o, Predicted NR 200456 Mortalityn (observed) events: Fair CRF, Assessment 1, 26-year follow-up <8.9%: 30.1 (37) 8.9–11.3%: 41.3 (44) 1.4–15.2%: 54.2 (52) 15.3–20.0%: 70.4 (65) >20.0%: 104.1 (102) Total: 300.1 (300) CRF+X, Assessment 1, 26-year follow-up <7.2%: 24.8 (27) 7.2–10.2%: 37.2 (37) 10.3–14.1%: 49.2 (47) 14.2–20.8%: 67.8 (69) >20.8%: 121.2 (120) Total: 300.2 (300) CRF, Assessment 2, 19-year follow-up <8.9%: 14.7 (16) 8.9–11.3%: 19.9 (19) 1.4–15.2%: 24.9 (21) 15.3–20.0%: 32.4 (38) >20.0%: 51.0 (49) Total: 142.9 (143) CRF+X, Assessment 2, 19-year follow-up <7.2%: 10.7 (10) 7.2–10.2%: 15.9 (18) 10.3–14.1%: 21.8 (27) 14.2–20.8%: 30.6 (23) >20.8%: 64.0 (65) Total: 142.9 (143) CRF, Assessment 1 with Insertion of Assessment 2 data for those who remained healthy (at 2), 19-year follow-up <8.9%: 15.8 (18) 8.9–11.3%: 21.4 (20) 1.4–15.2%: 27.1 (25) 15.3–20.0%: 36.0 (32) >20.0%: 58.6 (48) Total: 158.9 (143)

Screening for CVD Ri sk Wi th ECG 54 RTI–UNC EPC Table 5. Results of Included Studies for KQ 2 That Evaluated Exercise ECG

Fir s t Author, Year N (%) With Quality Outcome Outcome Discrimination Calibration Reclassification Erikssen, CRF+X, Assessment 1 with Insertion of 200456 Assessment 2 data for those who remained Fair healthy (at 2), 19-year follow-up (cont’d) <7.2%: 8.0 (12) 7.2–10.2%: 14.9 (15) 10.3–14.1%: 23.3 (30) 14.2–20.8%: 35.6 (31) >20.8%: 76.5 (55) Total: 158.3 (143) a Cardiac events were defined as a composite of cardiac death, nonfatal MI, and the need for coronary revascularization following the development of symptomatic CAD. b The 106 events included 17 cardiac death, 16 nonfatal MIs, and 73 coronary revascularizations. c Metabolic equivalents of task (peak exercise capacity was determined from the ETT to determine METs, and it was categorized as >8, 5 to 8, or <5). d Duke treadmill score (it was categorized as low, 5 or more, intermediate, 4 to -10, or high, -11 or less) e ETT is based on criteria for determining ischemia (separate from DTS or METs from the exercise test) f Appropriate use cohort was 824 patients (87% of the total cohort) considered acceptable candidates for functional testing on the basis of recent appropriate use criteria (i.e., intermediate to high FRS risk and/or chest pain symptoms). g Total coronary events included cardiac deaths, sudden deaths, acute MI, and stable or unstable angina. h Including 6 cardiac deaths, 13 stable or unstable angina events, and 15 nonfatal MI. Number of sudden deaths NR. i When adjusting for age, sex, current tobacco consumption, systolic blood pressure, total cholesterol, HDL cholesterol and diabetes (instead of 10-year Framingham risk of CHD), reported p was 0.10. j When adjusting for age, sex, current tobacco consumption, systolic blood pressure, total cholesterol, HDL cholesterol and diabetes (instead of 10-year Framingham risk of CHD), reported p was 0.03. k Definite coronary events included cardiac deaths, sudden deaths, acute MI, and stable or unstable angina. Revascularization (coronary artery bypass surgery or percutaneous coronary intervention) without clinical symptom was not considered as a coronary event. l Study reported 8 with sudden death or fatal MI, 24 with nonfatal MI, 15 with acute coronary syndromes, and 47 with stable angina. m All of the models considering the exercise test variable also included femoral bruit because it had been significant in adjusted HRs. The article provides data for FRS + femoral bruit also, showing that there was little to no change with its addition, e.g., AUROC 0.732 (same as for the model with FRS only). n Deaths caused by ischemic heart disease and sudden, unexpected deaths were classified as coronary deaths. Cardiovascular deaths also include deaths caused by stroke and ruptured aortic . o Exercise predictors: Physical fitness (cumulative work during exercise divided by body weight), maximal heart rate, systolic blood pressure at end of the first exercise load (100 W), and exercise ECG interpretation (ST-depression of at least 1.0 mm 0.08 s after the J-point regardless of ST-segment morphology)

Abbreviations: AUC=area under the curve; AUROC=area under the receiver operating characteristic curve; CHD=coronary heart disease; CRF=classical risk factors; DTS=Duke treadmill score; ECG=electrocardiogram; ETT=exercise treadmill testing; MET=metabolic equivalents of task; FRS=Framingham Risk Score; IDI=integrated discrimination improvement; KQ=key question; N=sample; NR=not reported; NRI=net reclassification improvement; SCORE=Systematic COronary Risk Evaluation.

Screening for CVD Ri sk Wi th ECG 55 RTI–UNC EPC Table 6. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 1

Table 6. Characteristics of I ncl u de d St u di es fo r KQ 2 That Evaluated RestingECG , Part 1 First Author, Year Model Type: Quality ECG Findings Evaluated Base model Cohort Source of Patients Country Auer et al, 201258 Majora and minorb 12-lead ECG Both types evaluated Health, Aging, Population-based cohort assessing U.S. Good abnormalities classified using Model development: and Body body composition, long-ter m the Minnesota Coding System (1) FRS variablesc and diabetes Composition conditions, and incident mobility (2) FRS variables only Study (Health limitation in an older adult cohort Published coefficient: FRS ABC) Study) (1997–98) Badheka et al, Major and Minor 12-lead ECG Published coefficient: FRS NHA NES-III Population-based survey to collect U.S. 201357 abnormalities classified using information on the health and Fair the Minnesota Coding Systemd nutrition of U.S. households (1988– 1994). Badheka et al, 12-lead ECG ST-T w ave Published coefficient: FRS NHA NES-III Population-based survey to collect U.S. 201363 abnormalities in lead aVR information on the health and Fair classified by the Minnesota nutrition of U.S. households (1988– Code 1994). Denes 200749 Major,e minor,f and incidentg 12- Published coefficient: FRS WHI Study Population-based study on common U.S. Fair lead ECG changes using the ( + causes of morbidity/mortality among Novacode criteria progestin trial) postmenopausal w omen (1993– 1998). Folsom 200364 LVH using a 12-lead ECG and Model development: A RIC Population-based study of 4 U.S. U.S. Fair the Cornell score FRS variablesh communities (1987–1989) Ishikaw a 201561 Prolonged corrected QT (QTc) Model development: The Jichi Government-sponsored screening to Japan Fair intervalsi and LVHj on 12-lead FRS variables plus alcohol intake Medical School clarify the risk factors for ECG and heart ratek Cohort cardio/cerebrovascular diseases in the general population (1992-1995) Jorgensen 201460 Major and Minor 12-lead ECG Model development: The The Copenhagen City Heart Study Denmark Fair abnormalities classified using FRS variablesl Copenhagen (1976–1978) Minnesota Code; also reported City Heart Study outcomes for some single ECG changesm Shah, 201659 ECG Risk Score including Both types evaluated NHA NES I Population-based survey to collect U.S. Fair frontal T axis, corrected QT (development information on the health and interval, T axis, heart rate, age, Published coefficient: FRSn and cohort) and nutrition; sex, age*sex interaction term PCE NHA NES III NHA NES I (1971–1975) and (selected from majoro and (validation NHA NES III (1988–1994) minorp abnormalities) Model development: cohort) FRS variables Tereshchenko, Resting 12-lead, P w ave Model development: A RIC Population-based study of 4 U.S. U.S. 201462 morphology (specifically FRS variablesq communities (1987–1989) Fair DTNPV 1r) a Criteria for major prevalent ECG abnormalities were any of the following: Q-QS wave abnormalities (MC 1-1 to 1-2-8); left ventricular hypertrophy (MC 3–1); Wolff-Parkinson- White syndrome (MC 6-4-1 or 6-4-2); complete bundle branch block or (MC 7-1-1, 7-2-1, 7–4, or 7–8); or (MC 8–3); or major ST -T changes (MC 4–1, 4–2, 5–1, and 5–2).

Screening for CVD Ri sk Wi th ECG 56 RTI–UNC EPC Table 6. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 1 b Criteria for minor prevalent ECG abnormalities were minor ST-T changes (MC 4– 3, 4-4, 5–3, and 5–4). Participants with both major and minor abnormalities were classified as having major abnormalities. Participants without minor or major ECG abnormalities were classified as having marginal or no abnormalities and their ECG was considered normal. c FRS variables were age, sex, total and HDL-C systolic blood pressure, and smoking. d Individuals with any of the following at baseline were considered to have ECG abnormalities: possible or probably MI, cardiac infarction/injury score of >=10, possible or probably left ventricular hypertrophy, any axis deviation, and any rhythm abnormalities other than sinus. e Criteria for major prevalent ECG abnormalities were any of the following: (1) atrial fibrillation or atrial flutter; (2) high-degree atrioventricular dissociation; (3) left bundle- branch block; (4) right bundle-branch block; (5) indeterminate conduction delay; (6) Qwave MI; (7) isolated ischemic abnormalities; (8) left ventricular hypertrophy with ST-T abnormalities; and (9) miscellaneous arrhythmias (e.g., supraventricular , ventricular preexcitation, ) with less than 5 participants being included in the analysis and not listed individually. Women with both major and minor abnormalities were classified as having major abnormalities. f Criteria for minor prevalent ECG abnormalities were any of the following: (1) first- and second-degree ; (2) borderline prolonged ventricular excitation; (3) prolonged ventricular repolarization; (4) isolated minor Q and ST-T abnormalities; (5) left ventricular hypertrophy without ST-T abnormalities; (6) left ; (7) frequent atrial or ventricular premature beats; and (8) fascicular blocks. g Criteria for incident ECG abnormalities were any of the following: (1) new atrial fibrillation or flutter; (2) new prolonged ventricular excitation; (3) new prolonged ventricular repolarization; (4) new left ventricular hypertrophy; (5) new Q-wave MI; and (6) new ischemic ST-T evolution. h model included age, race, total & HDL cholesterol, systolic blood pressure, use of antihypertensive medication, and smoking status i QTc determined by Bazett QTc intervals of ≥440 ms in men and ≥460 ms in women on a 12-lead ECG j LVH diagnosed with Cornell product of >=244 mVxms k model included age, sex, body mass index, current smoking, alcohol intake >20 g/d, systolic blood pressure, antihypertensive medication use, diabetes mellitus, hyperlipidemia, and heart rate. l Model included age, systolic blood pressure, total cholesterol, sex, current smoking, and diabet es m Reported outcomes for major or minor ECG changes, T wave changes, ventricular conduction delay, LVH, Q waves, ST depressions, resting heart rate n FRS model includes age, sex, systolic and diastolic blood pressure, diabetes, tobacco use, total and HDL-C levels, and use of antihypertensives o Major ECG abnormalities were defined based on Minnesota codes as follows: Major Q/QS waves (1.1, 1.2), ST depression (4.1, 4.2), negative T waves (5.1, 5.2), ventricular conduction defect (7.1, 7.2, or 7.4), atrial fibrillation/flutter (8.3), or ST elevation (9.2). p Minor ECG abnormalities were defined as having Minnesota codes for Minor Q waves (1.2.8 or 1.3), high R waves (3.1 or 3.3), minor ST changes (4.3 or 4.4), minor T wave changes (5.3 or 5.4), prolonged PR interval (6.3), RR’ in V1 or V2 (7.3 or 7.5), or left anterior fascicular block (7.7). q FRS components: age, gender, systolic blood pressure, diabetes, HDL and total cholesterol, smoking, and blood pressure-lowering therapy) r Deep terminal negativity of P wave in V1

Abbreviations: ARIC=Atherosclerosis Risk in Communities; ECG=electrocardiogram; FRS=Framingham Risk Score; HDL-C= high-density lipoprotein cholesterol; LVH=Left ventricular hypertrophy; NHANES-I= National Health and Nutrition Examination Survey-i; NHANES-III= National Health and Nutrition Examination Survey-III; PCE=pooled cohort equations; U.S.=United States; WHI=Women’s Health Initiative.

Screening for CVD Ri sk Wi th ECG 57 RTI–UNC EPC Table 7. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 2

Table 7. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 2 First Author, Year Sam ple Years of % With Mean % Non- Mean BMI % % Quality Size Follow -up % CV D Symptoms Age (SD) % F w hite CV Ris k (SD) % HTN DM Smokers Auer et al, 201258 2192 Median: 8.2 0 NR 73.5 (2.8) 55 41 FRS mean (SD): 27.4 (4.9) 57.3 13.3 Current: Good 12.6 (7.3a) 10.1 Mean intermediate Former: risk 43.6 Badheka et al, 6025 Mean 13 0 NRb 58.7 (13) 54 12 <5: 3391 (59%) 27.2 (5) 40 0 24 201357 5-10: 987 (17%) Fair 10-20: 854 (15%) >20: 497 (9%) Most low risk Badheka et al, 7928 Mean13.5 CA D: 9.8 NR 59.9 55 9.2 <5: 2625 (35%) 27.6 (5.5) 43.8 10.9 23.1 201363 MI: 5.4 (13.4) 5-10: 1221 (16%) Fair HF: 2.8 10-20: 1487 (20%) Stroke: 2.9 >20: 2176 (29%) Total: 15.4 Most low risk Denes 200749 1,264c Mean 5.6 0 NR 63 100 16 NR 28-29d 55-75 4 Pas t: 40 Fair (5.6-6.2) Current: 10 Folsom 200364 14,054 Median: 10.2 0 w ith NR Median: 57 NRe NR NR NR 10.7 NR Fair history of 55 (range CHD 45-64) Ishikaw a 201561 10,643 Mean 10.7 Unclear, but NR 55.4 62 NR <2.5: 4648 (55%) 23.1 (3.1) 33.9 3.6 22.6 Fair likely small (11.2) 2.5-5: 1819 (22%) %f >5: 1986 (23%) Most low stroke risk Jorgensen 201460 6991gi Median 11.9h 0 NR 70 (4) 59 NR FRSi 26 (4.3) NR 5 47 Fair 25.9-33.4 Mean high risk Shah, 201659 9969 Median: 18.8 0 NR Total: 53 Derivation: NHA NES ECG ris k NR Anti- 4.9; 34.3; Fair (derivation: (derivation), 55.3 11 equation scores: hypertensive 16.6 25.5j 3640, 10 (10.1) Validation: 9.02 (0.79); 8.96 use: 7.5; validation: (validation) 26.6 (0.86) 20.4 6329) Possibly low risk Tereshchenko, 15,375k Median 14 CHD: 5% NRl 54 (5.8) 55 27 <5: 12,463 (96%) 28 (5.5) 25 10 26 201462 HF:5% 5-20: 565 (4%) Fair MI: 4% >20: 21 (0.2%) Stroke: 2% Mos t low SCD ris k a Breakdown by FRS Categories by % 10-year risk was as follows: <5.0: 297 (13.6); 5.0-9.9: 525 (23.9); 10.0-19.9: 853 (38.9); ≥20.0: 517 (23.6) b St udy excluded t hose wit h self-reported chest pain suggestive of angina or leg pain suggestive of claudication c The number of participants shown here is the number in analyses eligible for our review. The authors used the 1,264 participants in the WHI blood subsample for the eligible analyses (the larger study included 14,749 participants, 7593 from the estrogen + progest in group and 7,156 from the placebo group). d When this table includes a range, it means that the data were not reported for the full sample, but were reported separately for subgroups

Screening for CVD Ri sk Wi th ECG 58 RTI–UNC EPC Table 7. Characteristics of Included Studies for KQ 2 That Evaluated Resting ECG, Part 2 e % nonwhite was NR for full sample, but t he aut hors report ed that 45% of diabetic participants were black. f Exclusion criteria listed pacemaker implantation, atrial fibrillation, advanced or complete atrioventricular block, dextrocardia, complete left or right bundle block, heart rate over 150 bpm, and history of stroke or MI, but did not address hist ory of CHD, TIA, angina, or PAD. g For sample with >=10 years follow-up (sample used to calculate discrimination and reclassification outcomes): 4923 had >=10 years of follow-up for the endpoint of fatal CVD, 5418 had >=10 years of follow-up for the endpoint of fatal and nonfatal CVD, and 6907 had >=10 years of follow-up for the endpoint of all-cause mortality. h Median for fatal CVD (primary outcome was 10.9); median for fatal or nonfatal CVD combined (secondary endpoint) was 9.8 years, and median for all-cause mortality was 11.9 years. i Baseline risk from the SCORE Risk Model: For participants with no ECG Changes: 13.1 (8.0–21.1); for those with ECG Changes: 18.3 (10.7–29.6) j When 2 numbers are present in these, they are for derivation cohort and the validation cohort k 13,049 CVD-free persons were in the reclassification analyses l T he st udy defined prevalent CHD to include those with symptoms of angina or claudication as well as those with diagnoses of CHD.

Abbreviations: AUC=area under the curve; CAD=coronary artery disease; CHD=coronary heart disease; CV=cardiovascular; CVD-cardiovascular disease; DM=diabetes mellitus; FRS=Framingham Risk Score; HF=heart failure; HTN=hypertension; MI=myocardial infarction; NHANES=National Health and Nutrition Examination Survey; NR=not reported; SCD=spontaneous cardiac death; SD=standard deviation.

Screening for CVD Ri sk Wi th ECG 59 RTI–UNC EPC Table 8. Summary of Evidence for Screening With ECG

No. of Studies, Key Question Study Summary of Main Findings (Including Limitations (Including Strength of and Topic Design N Consistency and Precision) Quality Reporting Bias) Evidence Applicability 1: Benefits of 2 RCTs 1,151 Neither study found a statistically significant 2 Fair Neither trial reached sample Low for no benefit Asymptomatic screening w ith reduction in events, including their primary size targets; stopped early of screening w ith adults ages 50 to ECG outcomes,a all-cause mortality, because of trouble recruiting. exercise ECG 75 years w ith cardiovascular-related mortality, MI, heart Not clear that 3.5 years of diabetes failure, or stroke. Findings w ere consistent follow up is sufficient. Masking Insufficient for undergoing and imprecise. of outcome assessors and resting ECG; no exercise ECG; amount of missing data NR in studies both trials 1.53 Reporting bias not enrolled high risk detected. populations 2: 5 cohort 9,582 Discrimination (k=3): small absolute 5 Fair Confidence intervals for Discrimination: Adults w ithout a Reclassification, studies improvement in AUC or C-statistics (0.02– calibration or discrimination NR Low for small history of CVD; calibration, and 0.03); none reported 95% CIs; 1 reported (k=5); mean duration of improvement mean age of discrimination for p=0.3 (no significant difference betw een follow up <10 yearsd (k=4), participants 50– exercise ECG models). Consistent; imprecise. reclassification NR (k=4); Calibration: 58 years; range unknow n masking of outcome Insufficient of females w as Calibration or performance (k=4 total; k=2 assessors (k=4); not reporting 0–38%; FRS base model): all 4 used different both discrimination and Reclassification: race/ethnicity NR metrics;b none reported figures such as calibration (k=3); model Insufficient in most (k=4); calibration plots;c k=3 reported improvement development studies (k=2); mean baseline w ith addition of exercise ECG variables; unclear handling and amount of FRS score w as mixed results for the 2 w ith FRS base missing data (k=2); the 1 study 10.8–12.3 in models. Inconsistent; imprecise. reporting NRI w as a model studies reporting development study, used risk it (k=3); Reclassification (k=1 model development categories of <6% vs. 6–20% intermediate risk, study, 988 participants): Total NRI 9.6% vs. >20% and may have on average (p=0.007); intermediate risk group NRI included many symptomatic 18.9% (p=0.01). Consistency unknow n; participants.e Reporting bias not imprecise. detected. 2: 9 cohort 66,407 Discrimination (k=7 total; k=4 FRS or PCE 8 Fair; Limited reporting on Discrimination: Adults w ithout a Reclassification, studies base model; k=4 multiple ECG changes): 1 Good assessment of symptoms; Low for very small history of CVD; calibration, and very small to small absolute improvement in unclear w hat proportion of to small mean age of discrimination for A UC or C-statistics (0.001–0.05); Few (k=3) participants w ere truly improvement participants 54– resting ECG reported w hether differences w ere asymptomatic. 73; majority w ere statistically significant. Consistent; Calibration: Low w omen in all imprecise. for improvement studies.

Screening for CVD Ri sk Wi th ECG 60 RTI–UNC EPC Table 8. Summary of Evidence for Screening With ECG

No. of Studies, Key Question Study Summary of Main Findings (Including Limitations (Including Strength of and Topic Design N Consistency and Precision) Quality Reporting Bias) Evidence Applicability 2: Calibration or performance (k=4 total; k=2 Masking of outcome assessors Reclassification: Range of Reclassification, FRS + major/minor ECG changes; k=1 FRS NR (k=8), confidence intervals Low for nonw hite calibration, and + specific T w ave change): none reported for calibration or discrimination improvement participants in discrimination for calibration plots; variety of metrics used; NR (k=5), not reporting those that resting ECG good calibration w ith addition of major/minor calibration (k=5), model reported (continued) changes (k=2) or T w ave amplitude in lead development studies (k=4), race/ethnicity aVR (k=1) to FRS. Poor calibration w ith amount of missing data NR (k=6) w as 9– addition of major/minor changes to FRS (k=2), and mean duration of 41%. Mean variables (k=1 model development of older follow up less than 10 years baseline risk adults 70–79). Consistent among studies (k=2). For reclassification, few ranging from low using published coefficients (k=3); studies (k=3) included a to high across imprecise. threshold betw een risk studies Reclassification (k=7 total w ith 59,123 categories corresponding to the participants; k=3 FRS or PCE + multiple recommendations for ECG changes; k=1 FRS + specific T w ave preventive medications (i.e., change). Overall, total NRIs (event; 7.5% or 10% 10-year risk). nonevent) range from 3.6% (2.7%; 0.6%) to 30% (17%; 19%) for studies using FRS or PCE base models (95% CIs rarely reported [Figure 7]).f Consistent in all show ing improved NRI, but inconsistent for estimates of NRI and outcomes assessed; consistency unknow n for specific risk categories because all studies used different risk categories; imprecise. 3: Harms of 1 RCT 520 One patient out of 12 (8.3%) undergoing 1 Fair Trial focused on assessing Insufficient Asymptomatic screening w ith revascularization procedures after positive benefits; did not reach sample adults ages 55 to ECG exercise treadmill tests in the DA DDY -D trial size target; not clear that mean 75 years w ith had a nonfatal acute MI 3 days after of 3.6 years of follow up is diabetes percutaneous revascularization and sufficient; masking of outcome undergoing underw ent a second percutaneous assessors NR and amount of screening w ith angioplasty.g Consistency unknow n (single missing data NR. Reporting exercise ECG study); imprecise. bias not detected. a For the primary composite outcomes, HRs were 1.00 (0.59 to 1.71) for a composite of death from all causes, nonfatal MI, nonfatal stroke, or heart failure requiring hospitalization or emergency service intervention, and 0.85 (0.39 to 1.84) for a composite of nonfatal MI or cardiac death. b Metrics included likelihood ratio test; Akaike information criteria (AIC), Brier’s score, and Hosmer-Lemeshow χ2; global χ2; and numbers of predicted and observed events. c One model development study provided a table of predicted and observed events for quintiles of risk.56 d The only study reporting longer followup covered 26 years, but it did not account for HDL in analyses.56 e 16.5% had atypical chest pain and participants were a subset of those having CACS and SPECT for “clinically indicated reasons.” 54

Screening for CVD Ri sk Wi th ECG 61 RTI–UNC EPC Table 8. Summary of Evidence for Screening With ECG f For multiple ECG changes (on resting 12-lead ECG), total NRIs (event NRIs; nonevent NRIs) for studies using any base model ranged from 1.9% (-0.2%; 0.6%) to 30% (17%; 19%). g The DADDY-D trial reported that 20/262 participants (7.6%) in the screened group had positive ETTs. Of those 20, 17 underwent coronary angiography (6.5% of the 262). Angiography revealed critical stenosis (not defined) in 71% (12/17), and all patients with critical stenosis underwent revascularization procedures (7 percutaneous, 5 surgical). The DYNAMIT trial (included in KQ 1) reported the number of some subsequent tests but did not report whether any of the tests or interventions resulted in harms; adverse events that occurred during followup were not recorded.52 Sixty eight of the 316 participants (21.5%) in the screened group had a definitely abnormal or an uncertain screening test (exercise test or SPECT) result. Of those, 38 underwent coronary angiography (12% of the 316 in the screened group) and 9 subsequently underwent coronary angioplasty (7/9 received stents) and 3 had coronary artery bypass graft.

Abbreviations: AIC=Akaike information criteria; AUC=area under the curve; CACS=coronary artery calcium score; CI=confidence interval; CVD=cardiovascular disease; DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; ECG=electrocardiogram; FRS=Framingham Risk Score; HDL=high-density lipoprotein; HR=hazard ratio; k=number of st udies; KQ=key question; MI=myocardial infarction; N=number; NR=not reported; NRI=net reclassification improvement; PCE=pooled cohort equations; RCT=randomized, controlled trial; SP ECT =single-photon emission computed tomography.

Screening for CVD Ri sk Wi th ECG 62 RTI–UNC EPC Table 9. Snapshot to Assess Net Benefit

Key Question Exercise ECG Resting ECG Considerations Benefits KQ 1: Benefits k=2; n=1,151 No evidence Despite enrolling high-risk persons with diabetes, neither trial of screening No statistically significant found benefit. But neither reached sample size targets; difference in primary composite stopped early because of trouble recruiting. Findings w ere outcomes, all-cause mortality, imprecise. Not clear that 3.5 years of follow up is sufficient. CV -related mortality, MI, heart Masking of outcome assessors and amount of missing data failure, or stroke at 3.5 years not reported in 1 trial. KQ 2: k=4; n=6,762 k=4; n=17,409 Preferred measures rarely reported. For resting ECG: none Calibration Mixed results (and all 4 used Improved calibration among reported calibration plots; limited reported on assessment of different metrics) studies using published coefficients symptoms; unclear w hat proportion of participants w ere truly of FRS (k=3). Poor calibration in 1 asymptomatic; majority of the resting ECG studies did not model development of older adults report calibration (k=5 out of 9); imprecise. ages 70–79 years. KQ 2: k=3; n=7,251 k=7; n=44,699 Overall, results w ere consistent but imprecise. Discrimination Small improvement Very small to small improvement For exercise ECG: 95% CIs w ere not reported (1 reported p=0.3, no significant difference betw een models). For resting ECG: k=4 FRS or PCE base model; few (k=3) reported w hether differences w ere statistically significant. KQ 2: k=1; n=988 k=7; n=59,123 Heterogeneity and applicability of risk thresholds; 95% CIs Reclassification Improvement (total NRI 9.6%, Improvement for studies using rarely reported. For exercise ECG: 1 model development p=0.007; intermediate risk group FRS or PCE base models (k=4): study that used risk categories of <6% vs. 6–20% vs. >20% NRI 18.9%, p=0.01) total NRIs (event; nonevent) range and may have included many symptomatic participants. For from 3.6% (2.7%; 0.6%) to 30% resting ECG: consistent in all show ing improved NRI but (17%; 19%) inconsistent for estimates of NRI and outcomes assessed; consistency unknow n for specific risk categories because all studies used different risk categories. Har m s KQ 3: k=1; n=520 No evidence Only 1 study eligible w ith 1 event reported. More information Screening 1/12 persons (8.3%) undergoing about potential harms is in the Discussion and contextual revascularization after positive question 2. exercise tests had a nonfatal acute MI 3 days after the procedure Abbreviations: CI=confidence interval; CV=cardiovascular; ECG=electrocardiogram; FRS=Framingham Risk Score; KQ=key question; MI=myocardial infarction; NRI=net reclassification improvement; PCE=pooled cohort equations.

Screening for CVD Ri sk Wi th ECG 63 RTI–UNC EPC Appendix A Ta ble 1. Recent Recommendations on Screening for CVD Using ECG

Organization Population Recommendation ACP, 20151 Asymptomatic low -risk adults Do not screen for cardiac disease in asymptomatic, low -risk adults w ith resting or stress ECG. A CC Low global risk; regardless of ECG Exercise ECG is rarely an appropriate option because Foundation/A HA /ASE/ interpretability and ability to of the lack of a clear benefit/risk advantage. A SNC/HFSA /HRS/S CA I/ exercise SCCT/SCMR/STS, Exercise ECG may be an appropriate option because of 20142 Intermediate global risk; ECG variable evidence or agreement regarding the interpretable and able to exercise benefit/risk ratio, potential benefit based on practice experience in the absence of evidence, and/or variability in the population. AAFP, 20123 Asymptomatic low -risk adults Do not order annual ECGs or any other cardiac screening. ACPM, 20114 General adult population Do not routinely screen the general adult population using resting or exercise ECG. A CC/A HA , 20105 Asymptomatic adults w ith A resting ECG is reasonable for cardiovascular risk hypertension or diabetes assessment.

Asymptomatic adults without A resting ECG may be considered for cardiovascular hypertension or diabetes risk assessment.

Intermediate-risk asymptomatic An exercise ECG may be considered for cardiovascular adults (including sedentary adults risk assessment, particularly w hen attention is paid to considering starting a vigorous non-ECG markers such as exercise capacity. exercise program) AHA, 20056 Asymptomatic adults There is insufficient evidence to recommend exercise testing as a routine screening modality. Abbreviations: AAFP=American Academy of Family Physicians; ACC=American College of Cardiology; ACP=American College of Physicians; ACPM=American College of Preventive Medicine; AHA=American Heart Association; ASA=American Stroke Association; ASE=American Society of Echocardiography; ASNC=American Society of Nuclear Cardiology; CVD=cardiovascular disease; ECG=electrocardiogram; HFSA=Heart Failure Society of America; HRS=Heart Rhythm Society; Royal College of Physicians of Edinburgh; SCAI=Society for Cardiovascular Angiography and Interventions; SCCT=Society of Cardiovascular Computed Tomography; SCMR=Society for Cardiovascular Magnetic Resonance; STS=Society of Thoracic Surgeons.

Screening for CVD Ri sk Wi th ECG 64 RTI–UNC EPC Appendix A Table 2. Characteristics of Available and Recommended Cardiovascular and Coronary Risk Assessment Models

Appendix A Ta bl e 2. Ch ar a ct er i st i cs of Av ai la bl e an d Rec om men de d C ar d io va scu l ar an d C or o nar y Ri sk As ses sme nt Mo del s

Risk Score and Recommending Risk Factors Included in the Time Horizon and Derivation and External Body Model Outcome Validation Cohorts Limitations ACC/AHA Pooled • Age 10-year risk Derivation Cohorts: Baseline exams for source cohorts Cohort Equation, • Sex A RIC, CHS, CA RDIA , conducted >25 years ago 7 2013 • Race/ethnicity First hard CVD event Framingham/Framingham No equations for Hispanics or Asians; lack • Treated or untreated SBP (nonfatal MI, CHD Offspring of large external datasets w ith needed 8 A CC/A HA • TC death, fatal or nonfatal covariate data to validate in these • HDL-C stroke) External Validation Cohorts: subpopulations • Current smoking REGA RDS, MESA, Contemporary Small numbers of events in validation Cohort (ARIC, Framingham/ cohorts, particularly in low er-risk groups • Diabetes Framingham Offspring), Possible overprediction across risk strata

Rotterdam Study, WHS, PHS, Other CVD risk factors WHI Observational Study evaluated but not includeda Framingham • Age 10-year risk Derivation Cohort: Not limited to “hard” outcomes 9 CV D, 2008 • Sex Framingham Baseline exams for source cohorts • TC Any CV D event conducted >40 years ago Canadian • HDL-C (coronary death, MI, External Validation Cohorts: Derivation cohort predominately w hite w ith Cardiovascular • coronary insufficiency, MESA, WHI Observational Study high proportion of smokers (~40%) 10 SBP Society • Antihypertensive medication angina), Possible overprediction, potentially higher use cerebrovascular events, among men • Smoking peripheral artery disease (intermittent • Diabetes claudication), and • b (Family history) congestive heart failure QRISK2, 200811 • Age 10-year risk Derivation Cohort: Not externally validated • Sex U.K. primary care database; 2/3 of Derivation cohort predominantly w hite 12 NICE • Race/ethnicity CVD event (angina, participants randomly allocated to Recording of family history of CHD possibly • Smoking status MI, stroke, TIA) derivation dataset and 1/3 not systematic • SBP assigned to validation dataset Tow nsend deprivation score specific to the • TC:HDL-C ratio United Kingdom • Body mass index External Validation Cohort: N/A • Family history of CHD in

first-degree relative age <60 years • Tow nsend deprivation score • Treated hypertension • Rheumatoid • Chronic • Diabetes • Atrial fibrillation

Screening for CVD Ri sk Wi th ECG 65 RTI–UNC EPC Appendix A Table 2. Characteristics of Available and Recommended Cardiovascular and Coronary Risk Assessment Models

Risk Score and Recommending Risk Factors Included in the Time Horizon and Derivation and External Body Model Outcome Validation Cohorts Limitations Reynolds, men, • Age 10-year risk Derivation Cohort: Derivation cohort predominately w hite 200813 • SBP PHS Derivation cohort health professionals; • Smoking CV D event (CV D health behaviors, access to health care, N/A • TC death, MI, stroke, External Validation Cohort: and SES may not be generalizable • HDL-C coronary MESA Data on blood pressure, obesity, and family • hs-CRP revascularization) history based on self-report • Parental history of MI at <60 Uncertain applicability in men <50 years old years of age and those w ith diabetes Reynolds, • Age 10-year risk Derivation Cohort: Derivation cohort predominately w hite 14 w omen, 2007 • SBP WHS; 2/3 of participants assigned Derivation cohort health professionals; • Smoking CV D ev ents (CV D to model derivation dataset and health behaviors, access to health care, N/A • TC death, MI, stroke, 1/3 assigned to validation dataset and SES may not be generalizable • HDL-C coronary Data on blood pressure, obesity, and family • hs-CRP revascularization) External Validation Cohorts: history based on self-report • Parental history of MI at <60 MESA, WHI Observational Study Possible underprediction years of age • HbA1c if diabetic ASSIGN, 200715 • TC 10-year risk Derivation Cohort: Not externally validated in the United States • HDL-C SHHEC Baseline exams for source cohort conducted 16 SIGN • SBP CV D ev ents (CV D >30 years ago • Smoking death, hospitalization External Validation Cohort: Social deprivation index specific to Scotland • Cigarettes per day f or CHD or U.K. general practice database High prevalence of smoking (~40%) and • Family history cerebrovascular family history (~20%) in source cohort • Diabetes disease, revascularization) • Index of social status/ deprivation A RIC, 200317 • Sex 10-year risk Derivation Cohort: Not externally validated • Rac e A RIC Baseline exams for source cohorts N/A • Cigarette smoking CHD ev ent (CHD conducted >25 years ago • TC death, MI, External Validation Cohort: Not limited to “hard” outcomes • HDL-C unrecognized MI N/A Race/ethnicity limited to blacks and w hites • SBP defined by Source cohort not inclusive of age <45 or electrocardiogram >65 years • Antihypertensive medication readings, or coronary use revascularization) • Diabetes

Other CV D risk f actors evaluated but not includedc

Screening for CVD Ri sk Wi th ECG 66 RTI–UNC EPC Appendix A Table 2. Characteristics of Available and Recommended Cardiovascular and Coronary Risk Assessment Models

Risk Score and Recommending Risk Factors Included in the Time Horizon and Derivation and External Body Model Outcome Validation Cohorts Limitations SCORE, 200318 • Age 10-year risk Derivation Cohort: Not externally validated in the United States • Sex Pooled dataset of population-based Baseline exams for source cohorts European Society • Smoking Fatal CVD event (MI, and occupational cohort studies conducted >25 years ago 19 of Cardiology • TC or TC:HDL ratio stroke, aortic from 12 European countries Diabetes not included as a risk factor • SBP ) because it w as not uniformly collected in • Smoking External Validation Cohorts: source cohort • High- and low -risk regions of Externally validated in European Europe cohorts (11 evaluation studies) PROCAM, 200220 • Age 10-year risk Derivation Cohort: Not externally validated in the United States • LDL-C Prospective German cohort of Baseline exams for source cohort conducted N/A • HDL-C CHD ev ent (sudden men >30 years ago • Triglycerides cardiac death, definite Excludes w omen and adults >65 years old • Smoking MI) External Validation Cohorts: Source cohort ~30% smokers • Diabetes Externally validated in several • Family history of MI at age European cohorts <60 years • SBP A TP III • Age 10-year risk Derivation Cohort: Baseline exams for source cohorts conducted modification of • Sex Framingham >40 years ago Wilson • TC Hard CHD (MI death, Derivation cohort predominately w hite w ith Framingham • HDL-C CHD death) External Validation Cohorts: high proportion of smokers (~40%) 21d model, 2002 • SBP A TP III: MESA , WHI Most validation cohorts have an upper age • Observational Study range of ages 64 or 74 years 21e Treatment for hypertension A TP III • Smoking Recent external validations of ATP III model Wilson: w ide range of cohorts in suggest substantial overestimation, United States, Europe, and particularly among men Australia, including ARIC, PHS, Older validations of the Wilson model show HHP, PR, SHS, CHS underprediction in high-risk groups (people w ho have diabetes, have a strong family history of premature CVD, reside in regions w ith high incidence, have low SES) and overprediction in low -risk groups (Japanese American men, Hispanic men, Native American w omen) a ACC/AHA recommends that if risk based treatment is uncertain using this tool, then consider one or more of the following: family history, hs-CRP, coronary artery calcium score, or ankle-brachial index. Do not use carotid intima-media thickness for risk assessment. No recommendation for or against use of , , microalbuminuria, and cardiorespiratory fitness. b Canadian Cardiovascular Society recommends a modified version of the model that includes family history of premature CHD.10 c Other CVD risk factors explored: age, body mass index, waist -to-hip ratio, sport activity index, forced expiratory volume, plasma , factor VII, factor VIII, von Willebrand factor, lipoprotein a, heart rate, Keys score, pack-years smoking, carotid intima-media thickness, fast ing triglycerides, apolipoprotein A, apolipoprotein B, albumin, white count, and creatinine.

Screening for CVD Ri sk Wi th ECG 67 RTI–UNC EPC Appendix A Table 2. Characteristics of Available and Recommended Cardiovascular and Coronary Risk Assessment Models d There are additional Framingham-based risk assessment models with variations in outcomes predicted and risk factors included.22-25 In this table we focused on models recommended by guideline bodies.9, 21 e Replaced by 2014 recommendations from the ACC/AHA.8

Abbreviations: ACC=American College of Cardiology; AHA=American Heart Association; ARIC=Atherosclerosis Risk in Communities Study; ATP III=Adult Treatment Panel III; CARDIA=Coronary Artery Risk Development in Young Adults; CHD=coronary heart disease; CHS=Cardiovascular Health Study; CVD=cardiovascular disease; HbA1c=glycat ed hemoglobin; HDL-C=low-density lipoprotein cholesterol; HHP=Honolulu Heart Program; hs-CRP=high-sensitivity C-reactive ; LDL-C=low-densit y lipoprotein cholesterol; MESA=Mult i-Ethnic Study of Atherosclerosis; MI=myocardial infarction; N/A=not applicable; NICE=National Institute for Health and Care Excellence; PHS=Physician’s Health Study; PR=Puert o Rico Heart Health Program; PROCAM=Prospective Cardiovascular Münster; REGARDS=Reasons for Geographic and Racial Differences in Stroke; SBP=systolic blood pressure; SCORE=Systematic Coronary Risk Evaluation; SES=; SHHEC=Scottish Heart Health Extended Cohort; SHS=St r o n g Hear t St udy ; SI GN=Scottish Intercollegiate Guidelines Network; TC=total cholesterol; TIA=transient ischemic attack; UK=Unit ed Kingdom; U.S.=United States; WHI=Women’s Health Initiative Observational Study; WHS=Women’s Health Study.

Screening for CVD Ri sk Wi th ECG 68 RTI–UNC EPC Appendix A. Contextual Questions

CQ 1a. For People in Each CVD Risk Category (or Strata), What Medications (i.e., Aspirin, Lipid-Lowering Therapy) Are Recommended?

Several organizations have recommendations for primary prevention of CVD, including the USPSTF, the ACC/AHA, and the American Academy of Family Physicians. European organizations have also weighed in on the subject. Most organizations consider aspirin and statin therapy for primary prevention, while some groups have also considered use.

Aspirin The USPSTF released their recommendation26 for the use of aspirin to prevent CVD in April 2016. They recommended initiating low-dose aspirin for adults ages 50 to 59 years with a 10-year CVD risk of 10 percent or greater and who are not at increased risk for bleeding, have a life expectancy of at least 10 years, and are willing to take daily aspirin for at least 10 years (B recommendation). The USPSTF recommended using the PCE for determination of 10-year risk. The USPSTF also has a C recommendation for adults ages 60 to 69 years (i.e., C recommendation indicating selectively offering this service based on professional judgment and patient preferences; at least moderate certainty that the net benefit is small). For adults less than age 50 years or 70 years or older, the USPSTF concluded there was insufficient evidence to assess the balance of benefits and harms (I statement).

Recommendations from the American Academy of Family Physicians about aspirin use for primary prevention have been consistent with those of the USPSTF. The American College of Chest Physicians suggests that patients older than 50 years without symptomatic CVD use low-dose aspirin for primary CVD prevention.27 Both the ACC/AHA and the American Stroke Association recommend low-dose aspirin for adults with 10-year CVD risk greater than 6 percent,28 while the American Diabetes Association29 recommends low-dose aspirin in patients with both type 1 and 2 diabetes who have 10-year CVD risk greater than 10 percent and are not at increased risk for bleeding. The ADA does not recommend aspirin therapy in men younger than 50 years or most women younger than 60 years who have low CVD risk because the risk for bleeding outweighs the potential benefits of aspirin treatment.

Statins In December 2016, the USPSTF30 recommended low- to moderate-dose statins for primary prevention for adults ages 40 to 75 years with one or more CVD risk factors and a 10-year CVD risk of 10 percent or greater (B recommendation). For adults with 10-year CVD risk between 7.5 and 10 percent, the USPSTF made a C recommendation, noting that the likelihood of benefit within this risk category was smaller but some adults might benefit. For adults older than 75 years, the USPSTF concluded the evidence was insufficient to assess the balance of benefits and harms (I statement).

The 2013 ACC/AHA guidelines31 recommend moderate- to high-intensity statin therapy for adults ages 40 to 75 years with LDL-C 70 to 189 mg/dL with a 10-year CVD risk greater than 7.5 percent as calculated by the PCE (strong recommendation) or if they have diabetes. For those with 10-year CVD risk of 5 percent to 7.5 percent, the ACC/AHA made a “weak recommendation” for using statins for primary prevention. For adults age 21 years or older with LDL-C greater than 190, they recommend statin therapy regardless of CVD risk (moderate recommendation).

Screening for CVD Ri sk Wi th ECG 69 RTI–UNC EPC Appendix A. Contextual Questions

The Canadian Cardiovascular Society recommends statins plus behavior modification for men age 40 years or older and women age 50 years or older without CVD risk factors and adults of any age with CVD risk factors who also have a 20 percent or greater 10-year CVD event risk (using FRS) or an LDL-C level of 135 to 19 0mg/dL and a 10 percent to 20 percent CVD event risk (based on the FRS).10 The recommended treatment strategy is treatment-to-target rather than by therapy dose (e.g., 50% reduction in LDL-C level). 10

The UK National Institute for Health and Care Excellence (NICE) recommends atorvastatin (20 mg) for primary prevention in adults (with or without type 2 diabetes) age 40 years or older with 10 percent or greater 10-year CVD risk (based on the QRISK2 tool).12, 32 They note that for adults age 85 years or older, statins may reduce the risk of nonfatal MI. They recommend statin treatment for adults with who are older than 40 years, have had diabetes more than 10 years, have nephropathy, or have other CVD risk factors.

Vitamin Supplements The USPSTF recommends against the use of beta-carotene and E supplementation for CVD prevention (D recommendation). For multivitamin, single-, or paired-nutrient supplements, the USPSTF concluded there was insufficient evidence to assess the balance of benefits and harms for CVD prevention (I statement). In addition to beta-carotene and , the AHA33 recommends against the use of vitamin supplements (e.g., ) and folic acid for primary CVD prevention.

Omega-3 Fatty Acids The AHA34 reported there were no RCTs to guide recommendations for the use of omega-3 fatty acids for primary CVD prevention in the general population. However, they found evidence there was no benefit for patients with or at risk for diabetes mellitus to prevent CVD. For patients with high CVD risk, the expert panel was split between recommending against omega-3 fatty acids use versus a weak recommendation that treatment in high-risk patients may be reasonable. NICE35 recommends against use of omega-3 fatty acids for primary CVD prevention.

CQ 1b. What Is the Fidelity to Prescribing and Taking the Recommended Medications?

According to the National Ambulatory and National Hospital Ambulatory Medical Care Surveys from 2005 to 2008,36 few patients were recommended to take aspirin in concordance with guideline recommendations. For example, in 2007 to 2008, among the population identified by the USPSTF to receive aspirin for prevention of CVD and stroke, it was recommended at only 16 percent of visits for males and 22 percent of visits for females.

In an analysis of the National Health and Nutrition Examination Survey (NHANES), 2011–2012,37 patients without CVD were classified into high and low risk based on FRSs Approximately 40 percent of the high-risk group and 26 percent of the low-risk group reported being told by their physician to take aspirin. Between 76 percent and 79 percent of patients advised to take aspirin reported complying. Using the same dataset, Malayala et al38 reported that about 35 percent of men ages 45 to 79 years who met USPSTF guideline recommendations for aspirin for primary prevention were advised by their providers to take aspirin and about 70 percent of that group reported compliance with the recommendation. Also using

Screening for CVD Ri sk Wi th ECG 70 RTI–UNC EPC Appendix A. Contextual Questions

the 2011–2012 NHANES dataset, Fiscella et al39 reported a slightly higher rate of aspirin recommendations for eligible women (i.e., ages 55 to 79 years), 42 percent, for primary CVD prevention.

The Reduction of Atherothrombosis for Continued Health (REACH)40 Registry compiled data on over 25,000 U.S. outpatients with atherothrombosis or atherosclerotic risk factors between 2003 2004. Approximately 25 percent of the registry enrollees were asymptomatic and comprised the primary prevention cohort (n=6,600). For primary CVD prevention, 62 percent of the cohort were taking at least one antiplatelet agent, most often aspirin, and 77 percent were receiving a statin.

CQ 2. What Are the Harms and Benefits of Revascularization Procedures for Adults Without Symptoms or a Prior Diagnosis of CVD?

Precise estimates of benefits and harms of revascularization for asymptomatic adults were not available. Available data come from studies with mostly symptomatic people. For example, population-based registries including people with symptoms estimate that the risk for any serious harms from angiography (which is often done at the same time as revascularization) is about 1.7 percent, including arrhythmia (0.4%), death (0.1%), stroke (0.07%), or MI (0.05%).41 After an abnormal screening ECG that is concerning for ischemia, some people without CHD would be sent for angiography without the possibility of benefit but would be subjected to potential harms. Even among a large sample undergoing elective angiography (about 400,000 participants) that was mostly symptomatic (70%), an estimated 39 percent had no CHD on angiography.42

One registry study published in JAMA Internal Medicine’s Less is More series reported data from a preoperative referral population (for noncardiac surgery) that describes data for a sample with a majority of asymptomatic participants (60%): the National Cardiovascular Data Registry CathPCI Registry. It is a large, national registry of patients undergoing diagnostic cardiac catheterizations and/or percutaneous coronary interventions (PCI) that captures about 85 percent of PCI procedures from approximately 1,400 U.S. hospitals. The registry43 contains data on nearly 195,000 patients who underwent preoperative evaluation prior to noncardiac surgery between July 2009 and December 2014. Approximately 60 percent of this cohort was reported to be clinically asymptomatic, although 58 percent had been taking medications within 2 weeks of the procedure; about 20 percent had atypical chest pain considered unlikely to be ischemic, and another 20 percent had stable angina. The sample excluded patients undergoing catheterization with suspected acute coronary syndrome, those who had unstable angina symptoms, and patients undergoing catheterization as part of a cardiac transplant evaluation.

The authors concluded that most patients undergoing diagnostic catheterization before noncardiac surgery are asymptomatic; the discovery of obstructive coronary artery disease is common; and although randomized, clinical trials have found no benefit in outcomes, revascularization is recommended in nearly half of these patients.

Obstructive disease was identified in 48 percent of the cohort overall, 40 percent of those who underwent diagnostic catheterization only, and 97 percent of those receiving PCI. Of patients with asymptomatic presentations, 48 percent were found to have obstructive disease. Approximately 16 percent of the cohort underwent PCI, and an additional 8 percent received CABG; revascularization was recommended in 23 percent of asymptomatic patients.

Screening for CVD Ri sk Wi th ECG 71 RTI–UNC EPC Appendix A. Contextual Questions

For the overall cohort, complications related to PCI were uncommon, including coronary artery (1.3%), periprocedural MI (1.7%), vascular complications (0.4%), stroke (0.1%), and renal failure (0.4%). Death occurred in 14 patients on the same day as the procedure, eight of which occurred in the catheterization lab. Bleeding events within 72 hours of the procedure occurred in 371 patients (1.3%), primarily at the procedure access site. Adverse events for asymptomatic patients were not separately reported.

Benefits of revascularization in asymptomatic adults is uncommonly reported. McFalls et al44 randomized patients scheduled for vascular surgery at 18 Veterans Affairs medical centers to receive preoperative coronary artery revascularization or no revascularization. Of these, 510 participants were randomized, and 240 participants underwent either PCI (n=141) or CABG (n=99). Notably, about half of all participants had prior CAD. After 2.7 years of followup, mortality was no different between groups (RR, 0.97; 95% CI, 0.69 to 1.36). In high-risk subgroups of participants, CABG did not confer a survival benefit. The vast majority of participants were taking beta-blockers, aspirin, ACE inhibitors, and statins up to 24 months after randomization, and their use did not differ between groups.

Screening for CVD Ri sk Wi th ECG 72 RTI–UNC EPC Appendix B1. Search Strategies

PubMed, 7/13/16 Search Que ry Items found

#1 Search ("Electrocardiography"[Mesh] OR electrocardiography OR EKG OR ECG OR "Exercise 256656 Test"[Mesh] OR (treadmill AND test) OR (treadmill AND ett))

#2 Search ("Myocardial Ischemia"[Mesh] OR "coronary heart disease"[tiab] OR "coronary 500580 disease"[tiab] OR "coronary disease"[mh] OR "coronary artery disease"[tiab] OR "Atherosclerosis"[Mesh] OR atherosclerosis[tiab])

#3 Search (#1 and #2) 67866

#4 Search (("Mass Screening"[Mesh] OR screen*[tiab])) 591681

#5 Search (#3 and #4) 1397

#6 Search (#3 and #4) Filters: Humans 1347

#7 Search (#3 and #4) Filters: Humans; English 1115

#8 Search (#3 and #4) Filters: Humans; English; Adult: 19+ years 843

#9 Search (((randomized[title/abstract] OR randomised[title/abstract]) AND controlled[title/abstract] 642180 AND trial[title/abstract]) OR (controlled[title/abstract] AND trial[title/abstract]) OR "Controlled Clinical Trial"[publication type] OR "Randomized Controlled Trial"[Publication Type] OR "Single- Blind Method"[MeSH] OR "Double-Blind Method"[MeSH] OR "Random Allocation"[MeSH])

#10 Search (#8 and #9) 81

#11 Search (#8 and #9) Filters: Humans 81

#12 Search (#8 and #9) Filters: Humans; English 81

#13 Search (#8 and #9) Filters: Publication date from 2009/01/01 to 2016/12/31; Humans; English 13

#14 Search (#3 and #4) Filters: Systematic Review s; Humans; English; Adult: 19+ years 18

#15 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Humans; English; Adult: 19+ 18 years

#16 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Publication date from 10 2009/01/01 to 2016/12/31; Humans; English; Adult: 19+ years

#17 Search ("adverse effects" [Subheading] OR "Long Term Adverse Effects"[Mesh] OR "Patient 3631548 Harm"[Mesh] OR "Mortality"[Mesh] OR "mortality" [Subheading] OR death[tiab] OR mortality[tiab] OR "medical errors"[mh] OR "iatrogenic disease"[mh] OR "false positive reactions"[mh] OR "Syncope"[Mesh] OR "Arrhythmias, Cardiac"[Mesh] OR "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Anxiety"[Mesh] OR labeling[tiab] OR labelling[tiab] OR "Coronary Angiography"[Mesh] OR "Myocardial Revascularization"[Mesh])

#18 Search (#8 and #17) 560

#19 Search ("Cohort Studies”[Mesh] OR “Epidemiologic Studies”[Mesh] OR “Follow -up 1903256 Studies”[Mesh] OR “prospective cohort” OR “prospective studies”[MeSH] OR (prospective*[All Fields] AND cohort[All Fields] AND (study[All Fields] OR studies[All Fields]))

#20 Search (#18 and #19) Publication date from 2009/01/01 to 2016/12/31 92

#21 Search (#8 and #17) Filters: Systematic Review s 13

#22 Search (#8 and #17) Filters: Systematic Review s; Meta-Analysis 13

#23 Search (#8 and #17) Filters: Systematic Review s; Meta-Analysis; Publication date from 8 2009/01/01 to 2016/12/31

#24 Search ("Coronary Angiography"[Mesh] OR "Myocardial Revascularization"[Mesh]) 121744

#25 Search (#2 and #24) 86047

#26 Search ("adverse effects" [Subheading] OR "Long Term Adverse Effects"[Mesh] OR "Patient 3574617 Harm"[Mesh] OR "Mortality"[Mesh] OR "mortality" [Subheading] OR death[tiab] OR mortality[tiab] OR "medical errors"[mh] OR "iatrogenic disease"[mh] OR "false positive reactions"[mh] OR "Syncope"[Mesh] OR "Arrhythmias, Cardiac"[Mesh] OR "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Anxiety"[Mesh] OR labeling[tiab] OR labelling[tiab])

#27 Search (#25 and #26) 49909

#28 Search (#25 and #26) Filters: Humans 49313

#29 Search (#25 and #26) Filters: Humans; English 42245

#30 Search (#25 and #26) Filters: Humans; English; Adult: 19+ years 32017

#31 Search (#25 and #26) Filters: Publication date from 2009/01/01 to 2016/12/31; Humans; English; 11463 Adult: 19+ years

#32 Search (#25 and #26) Filters: Systematic Review s; Publication date from 2009/01/01 to 266 2016/12/31; Humans; English; Adult: 19+ years

#33 Search (#25 and #26) Filters: Systematic Review s; Meta-Analysis; Publication date from 267 2009/01/01 to 2016/12/31; Humans; English; Adult: 19+ years

#34 Search (letter[pt] OR new spaper article[pt] OR editorial[pt] OR comment[pt]) 1511006

#35 Search (#33 not #34) 266

Screening for CVD Ri sk Wi th ECG 73 RTI–UNC EPC Appendix B1. Search Strategies

PubMed KQ 2 – Risk/Ha rms se a rch, 7/13/16 Search Que ry Items found

#1 Search ("Electrocardiography"[Mesh] OR electrocardiography OR EKG OR ECG OR "Exercise 256656 Test"[Mesh] OR (treadmill AND test) OR (treadmill AND ett))

#2 Search ("Mortality"[Mesh] OR "mortality"[Subheading] OR death[tiab] OR mortality[tiab] OR 1944815 "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Myocardial Ischemia"[Mesh] OR "Cardiovascular Diseases"[Mesh:NoExp] OR "Coronary Disease"[MeSH] OR "Coronary Disease"[mh] OR "coronary heart disease"[tiab] OR "coronary artery disease"[tiab] OR "coronary disease"[tiab] OR "Heart Failure"[Mesh] OR "heart failure"[tiab] OR "Stroke"[Mesh])

#3 Search (#1 and #2) 95297

#4 Search ("Risk"[Mesh:NoExp] OR "Logistic Models"[Mesh] OR "Risk Assessment"[Mesh] OR 1099100 "Risk Factors"[Mesh] OR "Predictive Value of Tests"[Mesh] OR "Kaplan-Meier Estimate"[Mesh] OR "risk prediction"[tiab] OR reclass*[tiab] OR Framingham[tiab] OR "risk score"[tiab] OR "risk scores"[tiab])

#5 Search (#3 and #4) 17436

#6 Search (#3 and #4) Filters: Humans 17238

#7 Search (#3 and #4) Filters: Humans; English 15162

#8 Search (#3 and #4) Filters: Humans; English; Adult: 19+ years 12007

#9 Search (#3 and #4) Filters: Publication date from 2009/01/01 to 2016/12/31; Humans; English; 4971 Adult: 19+ years

#10 Search (((randomized[title/abstract] OR randomised[title/abstract]) AND controlled[title/abstract] 642180 AND trial[title/abstract]) OR (controlled[title/abstract] AND trial[title/abstract]) OR "Controlled Clinical Trial"[publication type] OR "Randomized Controlled Trial"[Publication Type] OR "Single- Blind Method"[MeSH] OR "Double-Blind Method"[MeSH] OR "Random Allocation"[MeSH])

#11 Search (#9 and #10) 496

#12 Search ("Cohort Studies”[Mesh] OR “Epidemiologic Studies”[Mesh] OR “Follow -up 1903256 Studies”[Mesh] OR “prospective cohort” OR “prospective studies”[MeSH] OR (prospective*[All Fields] AND cohort[All Fields] AND (study[All Fields] OR studies[All Fields])))

#13 Search (#9 and #12) 3135

#14 Search (#3 and #4) Filters: Systematic Review s; Publication date from 2009/01/01 to 2016/12/31; 81 Humans; English; Adult: 19+ years

#15 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Publication date from 81 2009/01/01 to 2016/12/31; Humans; English; Adult: 19+ years

Screening for CVD Ri sk Wi th ECG 74 RTI–UNC EPC Appendix B1. Search Strategies

Cochrane Library Searches KQ 1 and KQ 3, 7/13/16 ID Search Hits #1 [mh Electrocardiography] or electrocardiography or EKG or ECG or [mh "Exercise Test"] or 20976 (treadmill and test) or (treadmill and ett) #2 [mh "Myocardial Ischemia"] or "coronary heart disease":ti,ab or "coronary disease":ti,ab or 32974 "coronary disease":kw or "coronary artery disease":ti,ab or [mh Atherosclerosis] or atherosclerosis:ti,ab #3 #1 and #2 5808 #4 [mh "Mass Screening"] or screen*:ti,ab 26321 #5 #3 and #4 Publication Year from 2009 to 2016, in Cochrane Review s (Review s and Protocols) and 13 Other Review s #6 ((controlled:ti or controlled:ab) and (trial:ti or trial:ab)) or "controlled clinical trial" or "randomized 574189 controlled trial":pt or "randomized controlled trial as topic":pt or "single-blind method":pt or "double- blind method":pt or "random allocation":pt #7 #5 and #6 9 #8 [mh /AE] or [mh "Long Term Adverse Effects"] or [mh "Patient Harm"] or [mh Mortality] or [mh /MO] 184054 or death:ti,ab or mortality:ti,ab or "medical errors":kw or "iatrogenic disease":kw or "false positive reactions":kw or [mh Syncope] or [mh "Arrhythmias, Cardiac"] or [mh "Myocardial Infarction"] or "heart attack":ti,ab or [mh Anxiety] or labeling:ti,ab or labelling:ti,ab or [mh "Coronary Angiography"] or [mh "Myocardial Revascularization"] #9 #5 and #8 12 #10 [mh "Cohort Studies"] or [mh "Epidemiologic Studies"] or [mh "Follow -up Studies"] or "prospective 139076 cohort" or [mh "prospective studies"] or (prospective* and cohort and (study or studies)) #11 #9 and #10 9 #12 [mh "Coronary Angiography"] or [mh "Myocardial Revascularization"] 11381 #13 #2 and #12 7785 #14 [mh /AE] or [mh "Long Term Adverse Effects"] or [mh "Patient Harm"] or [mh Mortality] or [mh /MO] 179573 or death:ti,ab or mortality:ti,ab or "medical errors":kw or "iatrogenic disease":kw or "false positive reactions":kw or [mh Syncope] or [mh "Arrhythmias, Cardiac"] or [mh "Myocardial Infarction"] or "heart attack":ti,ab or [mh Anxiety] or labeling:ti,ab or labelling:ti,ab #15 #13 and #14 Publication Year from 2009 to 2016, in Cochrane Review s (Review s and Protocols) 215 and Other Review s

Cochrane Library Searches KQ 2, Risk, 7/13/16 ID Search Hits #1 [mh Electrocardiography] or electrocardiography or EKG or ECG or [mh "Exercise Test"] or 20976 (treadmill and test) or (treadmill and ett) #2 [mh Mortality] or [mh /MO] or death:ti,ab or mortality:ti,ab or [mh "Myocardial Infarction"] or "heart 101003 attack":ti,ab or [mh "Myocardial Ischemia"] or [mh ^"Cardiovascular Diseases"] or [mh "Coronary Disease"] or "Coronary Disease":kw or "coronary heart disease":ti,ab or "coronary artery disease":ti,ab or "coronary disease":ti,ab or [mh "Heart Failure"] or "heart failure":ti,ab or [mh Stroke] #3 #1 and #2 8278 #4 [mh ^Risk] or [mh "Logistic Models"] or [mh "Risk Assessment"] or [mh "Risk Factors"] or [mh 44070 "Predictive Value of Tests"] or [mh "Kaplan-Meier Estimate"] or "risk prediction":ti,ab or reclass*:ti,ab or Framingham:ti,ab or "risk score":ti,ab or "risk scores":ti,ab #5 #3 and #4 1239 #6 #3 and #4 Publication Year from 2009 to 2016 539 #7 #6 in Cochrane Review s (Review s and Protocols) and Other Review s 33 #8 ((controlled:ti or controlled:ab) and (trial:ti or trial:ab)) or "controlled clinical trial" or "randomized 574189 controlled trial":pt or "randomized controlled trial as topic":pt or "single-blind method":pt or "double- blind method":pt or "random allocation":pt #9 #6 and #8 493 #10 [mh "Cohort Studies"] or [mh "Epidemiologic Studies"] or [mh "Follow -up Studies"] or "prospective 139076 cohort" or [mh "prospective studies"] or (prospective* and cohort and (study or studies)) #11 #6 and #10 281 #12 #9 or #11 Publication Year from 2009 to 2016 209

Screening for CVD Ri sk Wi th ECG 75 RTI–UNC EPC Appendix B1. Search Strategies

PubMed KQ 1 a nd KQ 3, 5/30/17 Items Search Que ry Found

#1 Search ("Electrocardiography"[Mesh] OR electrocardiography OR EKG OR ECG OR "Exercise 264065 Test"[Mesh] OR (treadmill AND test) OR (treadmill AND ett))

#2 Search ("Myocardial Ischemia"[Mesh] OR "coronary heart disease"[tiab] OR "coronary 518274 disease"[tiab] OR "coronary disease"[mh] OR "coronary artery disease"[tiab] OR "Atherosclerosis"[Mesh] OR atherosclerosis[tiab])

#3 Search (#1 and #2) 69195

#4 Search (("Mass Screening"[Mesh] OR screen*[tiab])) 630082

#5 Search (#3 and #4) 1445

#6 Search (#3 and #4) Filters: Humans 1390

#7 Search (#3 and #4) Filters: Humans; English 1156

#8 Search (#3 and #4) Filters: Humans; English; Adult: 19+ years 875

#9 Search (((randomized[title/abstract] OR randomised[title/abstract]) AND controlled[title/abstract] 672674 AND trial[title/abstract]) OR (controlled[title/abstract] AND trial[title/abstract]) OR "Controlled Clinical Trial"[publication type] OR "Randomized Controlled Trial"[Publication Type] OR "Single-Blind Method"[MeSH] OR "Double-Blind Method"[MeSH] OR "Random Allocation"[MeSH])

#10 Search (#8 and #9) 84

#11 Search (#8 and #9) Filters: Humans 84

#12 Search (#8 and #9) Filters: Humans; English 84

#13 Search (#8 and #9) Filters: Publication date from 2016/01/01 to 2017/12/31; Humans; English 4

#14 Search (#3 and #4) Filters: Systematic Review s; Humans; English; Adult: 19+ years 19

#15 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Humans; English; Adult: 19+ years 19

#16 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Publication date from 2009/01/01 1 to 2016/12/31; Humans; English; Adult: 19+ years

#17 Search ("adverse effects" [Subheading] OR "Long Term Adverse Effects"[Mesh] OR "Patient 3790291 Harm"[Mesh] OR "Mortality"[Mesh] OR "mortality" [Subheading] OR death[tiab] OR mortality[tiab] OR "medical errors"[mh] OR "iatrogenic disease"[mh] OR "false positive reactions"[mh] OR "Syncope"[Mesh] OR "Arrhythmias, Cardiac"[Mesh] OR "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Anxiety"[Mesh] OR labeling[tiab] OR labelling[tiab] OR "Coronary Angiography"[Mesh] OR "Myocardial Revascularization"[Mesh])

#18 Search (#8 and #17) 577

#19 Search ("Cohort Studies”[Mesh] OR “Epidemiologic Studies”[Mesh] OR “Follow -up Studies”[Mesh] 2023668 OR “prospective cohort” OR “prospective studies”[MeSH] OR (prospective*[All Fields] AND cohort[All Fields] AND (study[All Fields] OR studies[All Fields]))

#20 Search (#18 and #19) Publication date from 2016/01/01 to 2017/12/31 14

#21 Search (#8 and #17) Filters: Systematic Review s 13

#22 Search (#8 and #17) Filters: Systematic Review s; Meta-Analysis 13

#23 Search (#8 and #17) Filters: Systematic Review s; Meta-Analysis; Publication date from 2009/01/01 0 to 2016/12/31

#24 Search ("Coronary Angiography"[Mesh] OR "Myocardial Revascularization"[Mesh]) 125817

#25 Search (#2 and #24) 89049

#26 Search ("adverse effects" [Subheading] OR "Long Term Adverse Effects"[Mesh] OR "Patient 3731590 Harm"[Mesh] OR "Mortality"[Mesh] OR "mortality" [Subheading] OR death[tiab] OR mortality[tiab] OR "medical errors"[mh] OR "iatrogenic disease"[mh] OR "false positive reactions"[mh] OR "Syncope"[Mesh] OR "Arrhythmias, Cardiac"[Mesh] OR "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Anxiety"[Mesh] OR labeling[tiab] OR labelling[tiab])

#27 Search (#25 and #26) 51715

#28 Search (#25 and #26) Filters: Humans 51104

#29 Search (#25 and #26) Filters: Humans; English 43976

#30 Search (#25 and #26) Filters: Humans; English; Adult: 19+ years 33495

#31 Search (#25 and #26) Filters: Publication date from 2016/01/01 to 2017/12/31; Humans; English; 1221 Adult: 19+ years

#32 Search (#25 and #26) Filters: Systematic Review s; Publication date from 2016/01/01 to 22 2017/12/31; Humans; English; Adult: 19+ years

#33 Search (#25 and #26) Filters: Systematic Review s; Meta-Analysis; Publication date from 22 2016/01/01 to 2017/12/31; Humans; English; Adult: 19+ years

#34 Search (letter[pt] OR new spaper article[pt] OR editorial[pt] OR comment[pt]) 1566106

#35 Search (#33 not #34) 22

Screening for CVD Ri sk Wi th ECG 76 RTI–UNC EPC Appendix B1. Search Strategies

PubMed KQ 2 – Risk/Harms, 5/30/17 Items Search Que ry Found

#1 Search ("Electrocardiography"[Mesh] OR electrocardiography OR EKG OR ECG OR "Exercise 264065 Test"[Mesh] OR (treadmill AND test) OR (treadmill AND ett))

#2 Search ("Mortality"[Mesh] OR "mortality"[Subheading] OR death[tiab] OR mortality[tiab] OR 2043386 "Myocardial Infarction"[Mesh] OR "heart attack"[tiab] OR "Myocardial Ischemia"[Mesh] OR "Cardiovascular Diseases"[Mesh:NoExp] OR "Coronary Disease"[MeSH] OR "Coronary Disease"[mh] OR "coronary heart disease"[tiab] OR "coronary artery disease"[tiab] OR "coronary disease"[tiab] OR "Heart Failure"[Mesh] OR "heart failure"[tiab] OR "Stroke"[Mesh])

#3 Search (#1 and #2) 97776

#4 Search ("Risk"[Mesh:NoExp] OR "Logistic Models"[Mesh] OR "Risk Assessment"[Mesh] OR "Risk 1099100 Factors"[Mesh] OR "Predictive Value of Tests"[Mesh] OR "Kaplan-Meier Estimate"[Mesh] OR "risk prediction"[tiab] OR reclass*[tiab] OR Framingham[tiab] OR "risk score"[tiab] OR "risk scores"[tiab])

#5 Search (#3 and #4) 18197

#6 Search (#3 and #4) Filters: Humans 17983

#7 Search (#3 and #4) Filters: Humans; English 15883

#8 Search (#3 and #4) Filters: Humans; English; Adult: 19+ years 12604

#9 Search (#3 and #4) Filters: Publication date from 2016/01/01 to 2017/12/31; Humans; English; 509 Adult: 19+ years

#10 Search (((randomized[title/abstract] OR randomised[title/abstract]) AND controlled[title/abstract] 672674 AND trial[title/abstract]) OR (controlled[title/abstract] AND trial[title/abstract]) OR "Controlled Clinical Trial"[publication type] OR "Randomized Controlled Trial"[Publication Type] OR "Single-Blind Method"[MeSH] OR "Double-Blind Method"[MeSH] OR "Random Allocation"[MeSH])

#11 Search (#9 and #10) 51

#12 Search ("Cohort Studies”[Mesh] OR “Epidemiologic Studies”[Mesh] OR “Follow -up Studies”[Mesh] 2023668 OR “prospective cohort” OR “prospective studies”[MeSH] OR (prospective*[All Fields] AND cohort[All Fields] AND (study[All Fields] OR studies[All Fields])))

#13 Search (#9 and #12) 334

#14 Search (#3 and #4) Filters: Systematic Review s; Publication date from 2009/01/01 to 2016/12/31; 8 Humans; English; Adult: 19+ years

#15 Search (#3 and #4) Filters: Systematic Review s; Meta-Analysis; Publication date from 2016/01/01 8 to 2017/12/31; Humans; English; Adult: 19+ years

Screening for CVD Ri sk Wi th ECG 77 RTI–UNC EPC Appendix B1. Search Strategies

Cochrane Library Searches KQ 1 and KQ 3, 5/30/17 ID Search Hits #1 [mh Electrocardiography] or electrocardiography or EKG or ECG or [mh "Exercise Test"] or (treadmill and 22891 test) or (treadmill and ett) #2 [mh "Myocardial Ischemia"] or "coronary heart disease":ti,ab or "coronary disease":ti,ab or "coronary 35404 disease":kw or "coronary artery disease":ti,ab or [mh Atherosclerosis] or atherosclerosis:ti,ab #3 #1 and #2 6048 #4 [mh "Mass Screening"] or screen*:ti,ab 31391 #5 #3 and #4 Publication Year from 2016 to 2017, in Cochrane Review s (Review s and Protocols) and Other 7 Review s #6 ((controlled:ti or controlled:ab) and (trial:ti or trial:ab)) or "controlled clinical trial" or "randomized controlled 663873 trial":pt or "randomized controlled trial as topic":pt or "single-blind method":pt or "double-blind method":pt or "random allocation":pt #7 #5 and #6 6 #8 [mh /AE] or [mh "Long Term Adverse Effects"] or [mh "Patient Harm"] or [mh Mortality] or [mh /MO] or 201669 death:ti,ab or mortality:ti,ab or "medical errors":kw or "iatrogenic disease":kw or "false positive reactions":kw or [mh Syncope] or [mh "Arrhythmias, Cardiac"] or [mh "Myocardial Infarction"] or "heart attack":ti,ab or [mh Anxiety] or labeling:ti,ab or labelling:ti,ab or [mh "Coronary Angiography"] or [mh "Myocardial Revascularization"] #9 #5 and #8 6 #10 [mh "Cohort Studies"] or [mh "Epidemiologic Studies"] or [mh "Follow -up Studies"] or "prospective cohort" 150064 or [mh "prospective studies"] or (prospective* and cohort and (study or studies)) #11 #9 and #10 4 #12 [mh "Coronary Angiography"] or [mh "Myocardial Revascularization"] 11848 #13 #2 and #12 8118 #14 [mh /AE] or [mh "Long Term Adverse Effects"] or [mh "Patient Harm"] or [mh Mortality] or [mh /MO] or 197054 death:ti,ab or mortality:ti,ab or "medical errors":kw or "iatrogenic disease":kw or "false positive reactions":kw or [mh Syncope] or [mh "Arrhythmias, Cardiac"] or [mh "Myocardial Infarction"] or "heart attack":ti,ab or [mh Anxiety] or labeling:ti,ab or labelling:ti,ab #15 #13 and #14 Publication Year from 2016 to 2017, in Cochrane Review s (Review s and Protocols) and 4 Other Review s

Cochrane Library Searches KQ 2, Risk, 5/30/17 ID Search Hits #1 [mh Electrocardiography] or electrocardiography or EKG or ECG or [mh "Exercise Test"] or (treadmill and 22891 test) or (treadmill and ett) #2 [mh Mortality] or [mh /MO] or death:ti,ab or mortality:ti,ab or [mh "Myocardial Infarction"] or "heart 115133 attack":ti,ab or [mh "Myocardial Ischemia"] or [mh ^"Cardiovascular Diseases"] or [mh "Coronary Disease"] or "Coronary Disease":kw or "coronary heart disease":ti,ab or "coronary artery disease":ti,ab or "coronary disease":ti,ab or [mh "Heart Failure"] or "heart failure":ti,ab or [mh Stroke] #3 #1 and #2 8800 #4 [mh ^Risk] or [mh "Logistic Models"] or [mh "Risk Assessment"] or [mh "Risk Factors"] or [mh "Predictive 47503 Value of Tests"] or [mh "Kaplan-Meier Estimate"] or "risk prediction":ti,ab or reclass*:ti,ab or Framingham:ti,ab or "risk score":ti,ab or "risk scores":ti,ab #5 #3 and #4 1334 #6 #3 and #4 Publication Year from 2009 to 2016 55 #7 #6 in Cochrane Review s (Review s and Protocols) and Other Review s 2 #8 ((controlled:ti or controlled:ab) and (trial:ti or trial:ab)) or "controlled clinical trial" or "randomized controlled 663874 trial":pt or "randomized controlled trial as topic":pt or "single-blind method":pt or "double-blind method":pt or "random allocation":pt #9 #6 and #8 54 #10 [mh "Cohort Studies"] or [mh "Epidemiologic Studies"] or [mh "Follow -up Studies"] or "prospective cohort" 150064 or [mh "prospective studies"] or (prospective* and cohort and (study or studies)) #11 #6 and #10 27 #12 #9 or #11 Publication Year from 2016 to 2017 54

Screening for CVD Ri sk Wi th ECG 78 RTI–UNC EPC Appendix B2. Eligibility Criteria

Include Exclude Populations Adults age ≥18 years w ithout symptoms or a diagnosis of Persons w ith a history of atherosclerotic CVD; studies of mixed populations of asymptomatic and disease or symptoms suggesting coronary symptomatic persons are eligible if results are reported heart disease; children and adolescents separately for asymptomatic persons or <20% of the sample is symptomatica Screening tests Resting ECG, exercise ECG Radiology tests (e.g., thallium scan, scintigraphy, computed tomography), echocardiography, and vectorcardiographyb Comparisons All KQs: Screened vs. nonscreened groups (i.e., risk No comparison, nonconcordant historical stratification using ECG plus traditional risk factors vs. risk control, comparative studies w ith other stratification using traditional risk factors alone) novel risk factors (e.g., comparing ECG vs. KQ 2: CVD risk assessment models that include ECG C-reactive protein); studies that compare findings compared w ith those that do not the risk for subsequent events betw een KQ 3: For harms of subsequent procedures/interventions, persons w ith and w ithout ECG studies that compare the procedure/intervention to no abnormalities and report associations (e.g., procedure/intervention are also eligible. For studies prospective cohort studies that report reporting rates of harms from exercise ECG or hazard ratios for outcomes associated w ith subsequent procedures/interventions, large registries or baseline T-w ave abnormalities) multicenter studies w ithout a control group that report rates of harms for asymptomatic persons are also eligible. Outcomes KQ 1: All-cause mortality, cardiovascular mortality, and KQ 2: Studies assessing the association cardiovascular events (myocardial infarction, angina, betw een ECG findings and outcomes (e.g., stroke, congestive heart failure, composite cardiovascular w ith adjusted hazard ratios) outcomes) KQ 2: Reclassification, calibration (the degree to w hich predicted and observed risk estimates agree, goodness- of-fit statistics), and discrimination (C-statistic/area under the curve) KQ 3: Mortality, arrhythmia, cardiovascular events, or injuries from exercise ECG; anxiety; labeling; harms of subsequent procedures or interventions initiated as a result of screening (e.g., subsequent angiography or revascularization procedures resulting in harm) Study designs All KQs: Randomized, controlled trials or controlled All other designs, narrative review s, clinical trials systematic review sc case reports, case KQs 2, 3: Prospective cohort studies are also eligible series, editorials, letters, cross-sectional KQ 3: Well-designed large retrospective cohort studies studies and w ell-designed case-control studies (only for rare events) are also eligible Setting Studies performed in primary care or occupational Studies performed in specialty settings, medicine settings, studies that recruit participants from studies of patients undergoing the general population preoperative evaluation Country Studies conducted in countries categorized as “Very High” on the 2014 Human Development Index (as defined by the United Nations Development Program) Language English Non-English Study quality Good or fair Poor (according to design-specific USPSTF criteria) a The a priori plan for mixed populations was to include studies if the results were reported separately for asymptomatic persons or if less than 10 percent of the sample was symptomatic, and to systematically determine whether that approach would result in the exclusion of any studies in which 10 to 50 percent of the population was symptomatic that should be considered for inclusion. We only identified one such study (16.5% of participants had atypical chest pain45) and decided to include the study because of the uncertainty around the appropriate threshold (for proportion of symptomatic patients that would alter study findings and substantially limit their applicability for the review quest ions). b Vectorcardiography is a method of recording the magnitude and direction of the electrical forces that are generated by the heart by means of a continuous series of vectors that form curving lines around a central point. c We will not abstract data from systematic reviews and will not include them in the results, but we will conduct separate searches for systematic reviews and search the references lists of all potentially relevant systematic reviews to identify relevant primary studies that our electronic searches did not identify.

Screening for CVD Ri sk Wi th ECG 79 RTI–UNC EPC Appendix B3. U.S. Preventive Services Task Force Quality Rating Criteria

Randomized, Controlled Trials and Cohort Studies

Criteria • Initial assembly of comparable groups • Randomized, controlled trials (RCTs)—adequate randomization, including concealment and whether potential confounders were distributed equally among groups; cohort studies—consideration of potential confounders with either restriction or measurement for adjustment in the analysis; consideration of inception cohorts • Maintenance of comparable groups (includes attrition, crossovers, adherence, and contamination) • Important differential loss to followup or overall high loss to followup • Measurements: Equal, reliable, and valid (includes masking of outcome assessment) • Clear definition of interventions • Important outcomes considered • Analysis: Adjustment for potential confounders for cohort studies or intention-to-treat analysis for RCTs; for cluster RCTs, correction for correlation coefficient

Definition of Ratings Based on Above Criteria Good: Meets all criteria: Comparable groups are assembled initially and maintained throughout the study (followup ≥80 percent); reliable and valid measurement instruments are used and applied equally to the groups; interventions are spelled out clearly; important outcomes are considered; and appropriate attention is given to confounders in analysis. In addition, intention-to-treat analysis is used for RCTs. Fair: Studies will be graded “fair” if any or all of the following problems occur, without the important limitations noted in the “poor” category below: Generally comparable groups are assembled initially but some question remains on whether some (although not major) differences occurred in followup; measurement instruments are acceptable (although not the best) and generally applied equally; some but not all important outcomes are considered; and some but not all potential confounders are accounted for. Intention-to- treat analysis is lacking for RCTs. Poor: Studies will be graded “poor” if any of the following major limitations exist: Groups assembled initially are not close to being comparable or maintained throughout the study; unreliable or invalid measurement instruments are used or not applied equally among groups (including not masking outcome assessment); and key confounders are given little or no attention. Intention-to-treat analysis is lacking for RCTs. Sources: U.S. Preventive Services Task Force, Procedure Manual, Appendix VI https://www.uspreventiveservicestaskforce.org/Page/Name/methods-and-processes Harris et al, 200146

Screening for CVD Ri sk Wi th ECG 80 RTI–UNC EPC Appendix B3. U.S. Preventive Services Task Force Quality Rating Criteria

Specific Questions Used to Guide Assessment of Included Studies

Items used to assess quality of RCTs for KQs 1 and 3

1. Was randomization adequate? 2. Was allocation concealment adequate? 3. Were groups similar at baseline? 4. Was intervention fidelity adequate? 5. What w as the reported adherence to the intervention? 6. What w as the overall attrition? 7. What w as the differential attrition? 8. Did the study have differential attrition or overall high attrition raising concern for bias? 9. Did the study have cross-overs or contamination raising concern for bias? 10. Were outcome measurements equal, valid, and reliable? 11. Were patients masked? 12. Were providers masked? 13. Were outcome assessors masked? 14. Was the duration of follow up adequate to assess the outcome? 15. What w as the method used to handle missing data? 16. Did the study use acceptable statistical methods? 17. Quality Rating

Additional Items used to assess quality of RCTs that address harms, KQ3

1. Were harms prespecified and defined? 2. Were ascertainment techniques for harms adequately described? 3. Were ascertainment techniques for harms equal, valid, and reliable? 4. Was duration of follow -up adequate for harms assessment? 5. Quality Rating

Items used to assess quality of relevant studies reporting reclassification, calibration, and discrimination, KQ 2

1. Does study sample adequately capture the population of interest (participant eligibility and recruitment)? 2. Was there selective inclusion of participants in the model based on data availability? 3. If participants are from a treatment RCT, is treatment accounted for? 4. Enrolled consecutive patients or a random sample? 5. Were selection criteria clearly described? 6. Is a valid and reliable definition and method for measurement of the outcomes reported? 7. Was the same outcome definition (and method for measurement) used in all patients? 8. Were the outcomes assessed w ithout know ledge of the candidate predictors (i.e., blinded)? 9. Is a valid and reliable definition and method for measurement and classification of candidate predictor(s) (ECG and risk factors) reported? 10. Were predictors assessed blinded for the outcome, and for each other (if relevant)? 11. How w as the predictor of interest (ECG) handled in the modelling? 12. Number (%) participants w ith missing data (include predictors and outcomes) 13. Did the study have high attrition raising concern for bias? 14. How w as missing data handled? 15. Were multiple measures of performance used (e.g., global fit, discrimination, calibration, net reclassification)? 16. Were both calibration and discrimination measures reported? Were confidence intervals reported? 17. Were a priori cut points used for classification measures (e.g., sensitivity, specificity, predictive values, NRI)? 18. Was a bias-corrected NRI used? 19. If net reclassification w as assessed, were appropriate clinical thresholds used to reclassify risk? 20. Method used for testing model performance: development dataset only or separate external validation? 21. In w hat w ay w as the population a separate external validation from the FRS or PCE? 22. Was the FRS or PCE recalibrated in the population before ECG w as added to the model? 23. Quality

Screening for CVD Ri sk Wi th ECG 81 RTI–UNC EPC Appendix C. Excluded Studies

X1: Non-English X2: Ineligible population X3: Ineligible/no screening/treatment X4: Ineligible/no comparison X5: Ineligible/no outcome X6: Ineligible setting X7: Ineligible study design X8: Appears to meet all criteria but ineligible country X9: Appears to meet all criteria but abstract only X10: Poor quality

1. Summaries for patients. Adding 10.1161/strokeaha.109.567800. PMID: electrocardiography to medical history and 20167922. Exclusion Code: X5. physical examination for evaluation before 7. Agarwal SK, Simpson RJ, Jr., Rautaharju P, sports participation in college athletes. Ann et al. Relation of ventricular premature Intern Med. 2010 Mar 2;152(5):I13. doi: complexes to heart failure (from the 10.7326/0003-4819-152-5-201003020- Atherosclerosis Risk In Communities 00001. PMID: 20194228. Exclusion Code: [ARIC] Study). Am J Cardiol. 2012 Jan X7. 1;109(1):105-9. doi: 2. Abudiab M, Aijaz B, Konecny T, et al. Use 10.1016/j.amjcard.2011.08.009. PMID: of functional aerobic capacity based on 21945138. Exclusion Code: X5. stress testing to predict outcomes in normal, 8. Ahmed HM, Al-Mallah MH, McEvoy JW, overweight, and obese patients. Mayo Clin et al. Maximal exercise testing variables and Proc. 2013 Dec;88(12):1427-34. doi: 10-year survival: fitness risk score 10.1016/j.mayocp.2013.10.013. PMID: derivation from the FIT Project. Mayo Clin 24290116. Exclusion Code: X2. Proc. 2015 Mar;90(3):346-55. doi: 3. Acampa W, Petretta M, Evangelista L, et al. 10.1016/j.mayocp.2014.12.013. PMID: Myocardial perfusion imaging and risk 25744114. Exclusion Code: X2. classification for coronary heart disease in 9. Ahmed T, Steward JA, O'Mahony MS. diabetic patients. The IDIS study: a Dyspnoea and mortality in older people in prospective, multicentre trial. Eur J Nucl the community: a 10-year follow-up. Age Med Mol Imaging. 2012 Mar;39(3):387-95. . 2012 Jul;41(4):545-9. doi: doi: 10.1007/s00259-011-1983-x. PM ID: 10.1093/ageing/afs049. PMID: 22522776. 22109666. Exclusion Code: X3. Exclusion Code: X2. 4. Adabag AS, Grandits GA, Prineas RJ, et al. 10. Al Rifai M, Patel J, Hung RK, et al. Higher Relation of heart rate parameters during Fitness Is Strongly Protective in Patients exercise test to sudden death and all-cause with Family History of Heart Disease: The mortality in asymptomatic men. Am J FIT Project. Am J Med. 2017 Cardiol. 2008 May 15;101(10):1437-43. Mar;130(3):367-71. doi: doi: 10.1016/j.amjcard.2008.01.021. PMID: 10.1016/j.amjmed.2016.09.026. PMID: 18471455. Exclusion Code: X5. 27751899. Exclusion Code: X4. 5. Agarwal SK, Chao J, Peace F, et al. 11. Al Rifai M, Schneider AL, Alonso A, et al. Premature ventricular complexes on sRAGE, inflammation, and risk of atrial screening electrocardiogram and risk of fibrillation: results from the Atherosclerosis ischemic stroke. Stroke. 2015 Risk in Communities (ARIC) Study. J May;46(5):1365-7. doi: Diabetes Complications. 2015 10.1161/strokeaha.114.008447. PMID: Mar;29(2):180-5. doi: 25873602. Exclusion Code: X5. 10.1016/j.jdiacomp.2014.11.008. PMID: 6. Agarwal SK, Heiss G, Rautaharju PM, et al. 25499973. Exclusion Code: X3. Premature ventricular complexes and the 12. Aladin AI, Whelton SP, Al-Mallah MH, et risk of incident stroke: the Atherosclerosis al. Relation of resting heart rate to risk for Risk In Communities (ARIC) Study. Stroke. all-cause mortality by gender after 2010 Apr;41(4):588-93. doi: considering exercise capacity (the Henry

Screening for CVD Ri sk Wi th ECG 82 RTI–UNC EPC Appendix C. Excluded Studies

Ford exercise testing project). Am J Cardiol. status. Ann Noninvasive Electrocardiol. 2014 Dec 1;114(11):1701-6. doi: 2012 Jul;17(3):214-8. doi: 10.1111/j.1542- 10.1016/j.amjcard.2014.08.042. PMID: 474X.2012.00512.x. PMID: 22816540. 25439450. Exclusion Code: X2. Exclusion Code: X7. 13. Albayrak S, Ozhan H, Aslantas Y, et al. 21. Armstrong AW, Azizi S, Wu J, et al. Predictors of major adverse cardiovascular , electrocardiographic events; results of population based MELEN characteristics, and incidence of atrial study with prospective follow-up. Eur Rev fibrillation. Arch Dermatol Res. 2013 Med Pharmacol Sci. 2015 Apr;19(8):1446- Dec;305(10):891-7. doi: 10.1007/s00403- 51. PMID: 25967720. Exclusion Code: X5. 013-1419-5. PMID: 24166719. Exclusion 14. Al-Mallah MH, Keteyian SJ, Brawner CA, Code: X4. et al. Rationale and design of the Henry 22. Aro AL, Anttonen O, Tikkanen JT, et al. Ford Exercise Testing Project (the FIT Intraventricular conduction delay in a project). Clin Cardiol. 2014 Aug;37(8):456- standard 12-lead electrocardiogram as a 61. doi: 10.1002/clc.22302. PMID: predictor of mortality in the general 25138770. Exclusion Code: X2. population. Circ Arrhythm Electrophysiol. 15. Al-Zaiti SS, Carey MG. The Prevalence of 2011 Oct;4(5):704-10. doi: Clinical and Electrocardiographic Risk 10.1161/circep.111.963561. PMID: Factors of Cardiovascular Death Among 21841194. Exclusion Code: X5. On-duty Professional . J 23. Aro AL, Anttonen O, Tikkanen JT, et al. Cardiovasc Nurs. 2015 Sep-Oct;30(5):440- Prevalence and prognostic significance of T- 6. doi: 10.1097/jcn.0000000000000165. wave inversions in right precordial leads of PMID: 24874885. Exclusion Code: X7. a 12-lead electrocardiogram in the middle- 16. Al-Zaiti SS, Fallavollita JA, Wu YW, et al. aged subjects. Circulation. 2012 May Electrocardiogram-based predictors of 29;125(21):2572-7. doi: clinical outcomes: a meta-analysis of the 10.1161/circulationaha.112.098681. PMID: prognostic value of ventricular 22576982. Exclusion Code: X4. repolarization. Heart Lung. 2014 Nov- 24. Aro AL, Huikuri HV. Electrocardiographic Dec;43(6):516-26. doi: predictors of sudden cardiac death from a 10.1016/j.hrtlng.2014.05.004. PMID: large Finnish general population cohort. J 24988910. Exclusion Code: X7. Electrocardiol. 2013 Sep-Oct;46(5):434-8. 17. Andersen K, Rasmussen F, Held C, et al. doi: 10.1016/j.jelectrocard.2013.06.016. Exercise capacity and muscle strength and PMID: 23871401. Exclusion Code: X7. risk of vascular disease and arrhythmia in 25. Aro AL, Huikuri HV, Tikkanen JT, et al. 1.1 million young Swedish men: cohort QRS-T angle as a predictor of sudden study. Bmj. 2015;351:h4543. doi: cardiac death in a middle-aged general 10.1136/bmj.h4543. PMID: 26378015. population. Europace. 2012 Jun;14(6):872- Exclusion Code: X5. 6. doi: 10.1093/europace/eur393. PMID: 18. Anttila I, Nikus K, Nieminen T, et al. 22183749. Exclusion Code: X4. Prevalence and prognostic value of poor R- 26. Artero EG, Jackson AS, Sui X, et al. wave progression in standard resting Longitudinal algorithms to estimate electrocardiogram in a general adult cardiorespiratory fitness: associations with population. The Health 2000 Survey. Ann nonfatal cardiovascular disease and disease- Med. 2010 Mar;42(2):123-30. doi: specific mortality. J Am Coll Cardiol. 2014 10.3109/07853890903555334. PMID: Jun 3;63(21):2289-96. doi: 20166814. Exclusion Code: X5. 10.1016/j.jacc.2014.03.008. PMID: 19. Arad Y, Goodman KJ, Roth M, et al. 24703924. Exclusion Code: X3. Coronary , coronary disease risk 27. Asia Pacific Cohort Studies C, Barzi F, Patel factors, C-reactive protein, and A, et al. Cardiovascular risk prediction tools atherosclerotic cardiovascular disease for populations in Asia. J Epidemiol events: the St. Francis Heart Study. J Am Community Health. 2007 Feb;61(2):115-21. Coll Cardiol. 2005 Jul 5;46(1):158-65. doi: doi: 10.1136/jech.2005.044842. PMID: 10.1016/j.jacc.2005.02.088. PMID: 17234869. Exclusion Code: X3. 15992651. Exclusion Code: X3. 28. Assmann G, Cullen P, Schulte H. Simple 20. Arbel Y, Birati EY, Shapira I, et al. T-wave scoring scheme for calculating the risk of amplitude is related to physical fitness acute coronary events based on the 10-year

Screening for CVD Ri sk Wi th ECG 83 RTI–UNC EPC Appendix C. Excluded Studies

follow-up of the prospective cardiovascular 10.1161/01.CIR.0000143226.40607.71. Munster (PROCAM) study. Circulation. PMID: 15451778. Exclusion Code: X5. 2002 Jan 22;105(3):310-5. PMID: 36. Bandorski D, Bogossian H, Ecke A, et al. 11804985. Exclusion Code: X3. Evaluation of the prognostic value of 29. Assmann G, Schulte H, Cullen P, et al. electrocardiography parameters and heart Assessing risk of myocardial infarction and rhythm in patients with pulmonary stroke: new data from the Prospective hypertension. Cardiol J. 2016;23(4):465-72. Cardiovascular Munster (PROCAM) study. doi: 10.5603/CJ.a2016.0044. PMID: Eur J Clin Invest. 2007 Dec;37(12):925-32. 27367480. Exclusion Code: X5. doi: 10.1111/j.1365-2362.2007.01888.x. 37. Barmpouletos D, Stavens G, Ahlberg AW, PMID: 18036028. Exclusion Code: X3. et al. Duration and type of therapy for 30. Assmann G, Schulte H, Seedorf U. diabetes: impact on cardiac risk stratification Cardiovascular risk assessment in the with stress electrocardiographic-gated : results from the SPECT myocardial perfusion imaging. J Prospective Cardiovascular Munster Nucl Cardiol. 2010 Dec;17(6):1041-9. doi: (PROCAM) Study. Int J Obes (Lond). 2008 10.1007/s12350-010-9293-4. PMID: May;32 Suppl 2:S11-6. doi: 20963539. Exclusion Code: X3. 10.1038/ijo.2008.29. PMID: 18469834. 38. Bastuji-Garin S, Deverly A, Moyse D, et al. Exclusion Code: X3. The Framingham prediction rule is not valid 31. Bachmann TN, Skov MW, Rasmussen PV, in a European population of treated et al. Electrocardiographic Tpeak-Tend hypertensive patients. J Hypertens. 2002 interval and risk of cardiovascular morbidity Oct;20(10):1973-80. PMID: 12359975. and mortality: Results from the Copenhagen Exclusion Code: X3. ECG study. Heart Rhythm. 2016 39. Bates RE, Omer M, Abdelmoneim SS, et al. Apr;13(4):915-24. doi: Impact of Stress Testing for Coronary 10.1016/j.hrthm.2015.12.027. PMID: Artery Disease Screening in Asymptomatic 26707793. Exclusion Code: X5. Patients With Diabetes Mellitus: A 32. Badheka AO, Rathod A, Marzouka GR, et Community-Based Study in Olmsted al. Isolated nonspecific ST-segment and T- County, Minnesota. Mayo Clin Proc. 2016 wave abnormalities in a cross-sectional Nov;91(11):1535-44. doi: United States population and Mortality 10.1016/j.mayocp.2016.07.013. PMID: (from NHANES III). Am J Cardiol. 2012 27720456. Exclusion Code: X4. Aug 15;110(4):521-5. doi: 40. Bauters C, Lemesle G. Screening for 10.1016/j.amjcard.2012.04.023. PMID: asymptomatic coronary artery disease in 22608358. Exclusion Code: X4. patients with diabetes mellitus: A systematic 33. Badheka AO, Singh V, Patel NJ, et al. QRS review and meta-analysis of randomized duration on electrocardiography and trials. BMC Cardiovasc Disord. 2016 May cardiovascular mortality (from the National 10;16:90. doi: 10.1186/s12872-016-0256-9. Health and Nutrition Examination Survey- PMID: 27165687. Exclusion Code: X7. III). Am J Cardiol. 2013 Sep 1;112(5):671- 41. Becker A, Leber A, Becker C, et al. 7. doi: 10.1016/j.amjcard.2013.04.040. Predictive value of coronary PMID: 23726176. Exclusion Code: X2. for future cardiac events in asymptomatic 34. Bakhoya VN, Kurl S, Laukkanen JA. T- individuals. Am Heart J. 2008 wave inversion on electrocardiogram is Jan;155(1):154-60. doi: related to the risk of acute coronary 10.1016/j.ahj.2007.08.024. PMID: syndrome in the general population. Eur J 18082507. Exclusion Code: X3. Prev Cardiol. 2014 Apr;21(4):500-6. doi: 42. Becker A, Leber AW, Becker C, et al. 10.1177/2047487312460022. PMID: Predictive value of coronary calcifications 22952285. Exclusion Code: X2. for future cardiac events in asymptomatic 35. Balady GJ, Larson MG, Vasan RS, et al. patients with diabetes mellitus: a prospective Usefulness of exercise testing in the study in 716 patients over 8 years. BMC prediction of coronary disease risk among Cardiovasc Disord. 2008 Oct 10;8:27. doi: asymptomatic persons as a function of the 10.1186/1471-2261-8-27. PMID: 18847481. Framingham risk score. Circulation. 2004 Exclusion Code: X3. Oct 5;110(14):1920-5. doi: 43. Bem D, Lordkipanidze M, Hodgkinson J, et al. The Effects of Different Aspirin Dosing

Screening for CVD Ri sk Wi th ECG 84 RTI–UNC EPC Appendix C. Excluded Studies

Frequencies and the Timing of Aspirin associated with increased stroke risk. Stroke. Intake in Primary and Secondary Prevention 2011 Nov;42(11):3196-201. doi: of Cardiovascular Disease: A Systematic 10.1161/strokeaha.110.607697. PMID: Review. Clin Pharmacol Ther. 2016 21921280. Exclusion Code: X3. Nov;100(5):500-12. doi: 10.1002/cpt.438. 52. Blair SN, Kampert JB, Kohl HW, 3rd, et al. PMID: 27449968. Exclusion Code: X2. Influences of cardiorespiratory fitness and 44. Bernard S, Serusclat A, Targe F, et al. other precursors on cardiovascular disease Incremental predictive value of carotid and all-cause mortality in men and women. ultrasonography in the assessment of JAMA. 1996 Jul 17;276(3):205-10. PMID: coronary risk in a cohort of asymptomatic 8667564. Exclusion Code: X5. type 2 diabetic subjects. Diabetes Care. 53. Bodegard J, Erikssen G, Bjornholt JV, et al. 2005 May;28(5):1158-62. PMID: 15855582. Symptom-limited exercise testing, ST Exclusion Code: X3. depressions and long-term coronary heart 45. Berry JD, Lloyd-Jones DM, Garside DB, et disease mortality in apparently healthy al. Framingham risk score and prediction of middle-aged men. Eur J Cardiovasc Prev coronary heart disease death in young men. Rehabil. 2004 Aug;11(4):320-7. PMID: Am Heart J. 2007 Jul;154(1):80-6. doi: 15292766. Exclusion Code: X5. 10.1016/j.ahj.2007.03.042. PMID: 54. Boman K, Gerdts E, Wachtell K, et al. 17584558. Exclusion Code: X3. Exercise and cardiovascular outcomes in 46. Berry JD, Willis B, Gupta S, et al. Lifetime hypertensive patients in relation to structure risks for cardiovascular disease mortality by and function of left ventricular hypertrophy: cardiorespiratory fitness levels measured at the LIFE study. Eur J Cardiovasc Prev ages 45, 55, and 65 years in men. The Rehabil. 2009 Apr;16(2):242-8. doi: Cooper Center Longitudinal Study. J Am 10.1097/HJR.0b013e328329560e. PMID: Coll Cardiol. 2011 Apr 12;57(15):1604-10. 19369828. Exclusion Code: X2. doi: 10.1016/j.jacc.2010.10.056. PMID: 55. Bouzas-Mosquera A, Peteiro J, Alvarez- 21474041. Exclusion Code: X5. Garcia N, et al. Prediction of mortality and 47. Bhatia RS, Bouck Z, Ivers NM, et al. major cardiac events by exercise Electrocardiograms in Low-Risk Patients echocardiography in patients with normal Undergoing An Annual Health Examination. exercise electrocardiographic testing. J Am JAMA Intern Med. 2017 Jul 10doi: Coll Cardiol. 2009 May 26;53(21):1981-90. 10.1001/jamainternmed.2017.2649. PMID: doi: 10.1016/j.jacc.2009.01.067. PMID: 28692719. Exclusion Code: X5. 19460612. Exclusion Code: X2. 48. Bhopal R, Fischbacher C, Vartiainen E, et 56. Brindle P, Emberson J, Lampe F, et al. al. Predicted and observed cardiovascular Predictive accuracy of the Framingham disease in South Asians: application of coronary risk score in British men: FINRISK, Framingham and SCORE models prospective cohort study. BMJ. 2003 Nov to Newcastle Heart Project data. J Public 29;327(7426):1267. doi: Health (Oxf). 2005 Mar;27(1):93-100. doi: 10.1136/bmj.327.7426.1267. PMID: 10.1093/pubmed/fdh202. PMID: 15749725. 14644971. Exclusion Code: X3. Exclusion Code: X3. 57. Brindle PM, McConnachie A, Upton MN, et 49. Bianco HT, Izar MC, Fonseca HA, et al. al. The accuracy of the Framingham risk- Relevance of target-organ lesions as score in different socioeconomic groups: a predictors of mortality in patients with prospective study. Br J Gen Pract. 2005 diabetes mellitus. Arq Bras Cardiol. 2014 Nov;55(520):838-45. PMID: 16281999. Oct;103(4):272-81. PMID: 25098376. Exclusion Code: X3. Exclusion Code: X5. 58. Brown DW, Giles WH, Croft JB. Left 50. Binici Z, Intzilakis T, Nielsen OW, et al. ventricular hypertrophy as a predictor of Excessive supraventricular ectopic activity coronary heart disease mortality and the and increased risk of atrial fibrillation and effect of hypertension. Am Heart J. 2000 stroke. Circulation. 2010 May Dec;140(6):848-56. doi: 4;121(17):1904-11. doi: 10.1067/mhj.2000.111112. PMID: 10.1161/circulationaha.109.874982. PMID: 11099987. Exclusion Code: X5. 20404258. Exclusion Code: X3. 59. Buonacera A, Boukhris M, Tomasello SD, 51. Binici Z, Mouridsen MR, Kober L, et al. et al. Impact of left ventricular remodeling Decreased nighttime heart rate variability is and renal function on 24h-ECG recordings

Screening for CVD Ri sk Wi th ECG 85 RTI–UNC EPC Appendix C. Excluded Studies

and cardiovascular outcome in elderly 67. Chan AK, Ilias-Khan NA, Xian H, et al. hypertensive patients. Eur J Intern Med. Arm exercise stress perfusion imaging 2016 Apr;29:71-7. doi: predicts clinical outcome. J Appl Physiol 10.1016/j.ejim.2016.01.001. PMID: (1985). 2011 Dec;111(6):1546-53. doi: 26781517. Exclusion Code: X2. 10.1152/japplphysiol.00725.2011. PMID: 60. Buyken AE, von Eckardstein A, Schulte H, 21852405. Exclusion Code: X2. et al. Type 2 diabetes mellitus and risk of 68. Chauhan K, Ackerman MJ, Crowson CS, et coronary heart disease: results of the 10-year al. Population-based study of QT interval follow-up of the PROCAM study. Eur J prolongation in patients with rheumatoid Cardiovasc Prev Rehabil. 2007 arthritis. Clin Exp Rheumatol. 2015 Jan- Apr;14(2):230-6. doi: Feb;33(1):84-9. PMID: 25572282. 10.1097/HJR.0b013e3280142037. PMID: Exclusion Code: X5. 17446801. Exclusion Code: X4. 69. Cheng S, Keyes MJ, Larson MG, et al. 61. Camhi SM, Katzmarzyk PT, Broyles S, et Long-term outcomes in individuals with al. Association of metabolic risk with prolonged PR interval or first-degree longitudinal physical activity and fitness: atrioventricular block. Jama. 2009 Jun coronary artery risk development in young 24;301(24):2571-7. doi: adults (CARDIA). Metab Syndr Relat 10.1001/jama.2009.888. PMID: 19549974. Disord. 2013 Jun;11(3):195-204. doi: Exclusion Code: X5. 10.1089/met.2012.0120. PMID: 23438155. 70. Cheng S, Larson MG, Keyes MJ, et al. Exclusion Code: X3. Relation of QRS width in healthy persons to 62. Cardoso CR, Leite NC, Salles GF. Factors risk of future permanent pacemaker as s ociated with abnormal T-wave axis and implantation. Am J Cardiol. 2010 Sep increased QRS-T angle in type 2 diabetes. 1;106(5):668-72. doi: Acta Diabetol. 2013 Dec;50(6):919-25. doi: 10.1016/j.amjcard.2010.04.021. PMID: 10.1007/s00592-013-0483-9. PMID: 20723643. Exclusion Code: X5. 23744129. Exclusion Code: X2. 71. Cheriyath P, He F, Peters I, et al. Relation of 63. Carlson N, Dixen U, Marott JL, et al. atrial and/or ventricular premature Predictive value of casual ECG-based complexes on a two-minute rhythm strip to resting heart rate compared with resting the risk of sudden cardiac death (the heart rate obtained from Holter recording. Atherosclerosis Risk in Communities Scand J Clin Lab Invest. 2014 [ARIC] study). Am J Cardiol. 2011 Jan Mar;74(2):163-9. doi: 15;107(2):151-5. doi: 10.3109/00365513.2013.867531. PMID: 10.1016/j.amjcard.2010.09.002. PMID: 24329133. Exclusion Code: X2. 21211594. Exclusion Code: X5. 64. Carrington MJ, Jennings GL, Clark RA, et 72. Chisalita SI, Dahlstrom U, Arnqvist HJ, et al. Assessing cardiovascular risk in regional al. Proinsulin and IGFBP-1 predicts areas: the Healthy Beyond City mortality in an elderly population. Int J Limits program. BMC Health Serv Res. Cardiol. 2014 Jun 15;174(2):260-7. doi: 2012;12:296. doi: 10.1186/1472-6963-12- 10.1016/j.ijcard.2014.03.171. PMID: 296. PMID: 22943553. Exclusion Code: X2. 24794551. Exclusion Code: X3. 65. Cederholm J, Eeg-Olofsson K, Eliasson B, 73. Chou R, Arora B, Dana T, et al. Screening et al. Risk prediction of cardiovascular asymptomatic adults with resting or exercise disease in type 2 diabetes: a risk equation electrocardiography: a review of the from the Swedish National Diabetes evidence for the U.S. Preventive Services Register. Diabetes Care. 2008 Task Force. Ann Intern Med. 2011 Sep Oct;31(10):2038-43. doi: 10.2337/dc08- 20;155(6):375-85. doi: 10.7326/0003-4819- 0662. PMID: 18591403. Exclusion Code: 155-6-201109200-00006. PMID: 21930855. X3. Exc lusion Code: X7. 66. Chambless LE, Heiss G, Shahar E, et al. 74. Cicero AF, Rosticci M, Tocci G, et al. Prediction of ischemic stroke risk in the Serum and other short-term Atherosclerosis Risk in Communities Study. predictors of electrocardiographic alterations Am J Epidemiol. 2004 Aug 1;160(3):259-69. in the Brisighella Heart Study cohort. Eur J doi: 10.1093/aje/kwh189. PMID: 15257999. Intern Med. 2015 May;26(4):255-8. doi: Exclusion Code: X4. 10.1016/j.ejim.2015.02.007. PMID: 25708168. Exclusion Code: X5.

Screening for CVD Ri sk Wi th ECG 86 RTI–UNC EPC Appendix C. Excluded Studies

75. Cilli A, Batmaz F, Demir I, et al. The QRS complex: the ARIC Study with 13 diagnostic yield of exercise stress testing as years of follow-up. Circulation. 2003 Oct a screening tool for subclinical coronary 21;108(16):1985-9. doi: artery disease in patients with moderate to 10.1161/01.CIR.0000095027.28753.9D. severe obstructive apnea. J Clin Sleep PMID: 14517173. Exclusion Code: X5. Med. 2011 Feb 15;7(1):25-9. PMID: 84. Cuddy TE, Tate RB. Sudden unexpected 21344047. Exclusion Code: X7. cardiac death as a function of time since the 76. Cintra FD, Tufik S, Paola A, et al. detection of electrocardiographic and Cardiovascular profile in patients with clinical risk factors in apparently healthy obstructive . Arq Bras Cardiol. men: the Manitoba Follow-Up Study, 1948 2011 Apr;96(4):293-9. PMID: 21437515. to 2004. Can J Cardiol. 2006 Mar Exclusion Code: X5. 1;22(3):205-11. PMID: 16520850. 77. Cirillo M, Terradura-Vagnarelli O, Mancini Exclusion Code: X5. M, et al. Cohort profile: The Gubbio 85. D'Agostino RB, Sr., Grundy S, Sullivan LM, Population Study. Int J Epidemiol. 2014 et al. Validation of the Framingham Jun;43(3):713-20. doi: 10.1093/ije/dyt025. coronary heart disease prediction scores: PMID: 23543599. Exclusion Code: X5. results of a multiple ethnic groups 78. Cole CR, Foody JM, Blackstone EH, et al. investigation. JAMA. 2001 Jul Heart rate recovery after submaximal 11;286(2):180-7. PMID: 11448281. exercise testing as a predictor of mortality in Exclusion Code: X3. a cardiovascularly healthy cohort. Ann 86. D'Agostino RB, Sr., Vasan RS, Pencina MJ, Intern Med. 2000 Apr 4;132(7):552-5. et al. General cardiovascular risk profile for PMID: 10744592. Exclusion Code: X4. use in primary care: the Framingham Heart 79. Conen D, Schon T, Aeschbacher S, et al. Study. Circulation. 2008 Feb 12;117(6):743- Genetic and phenotypic determinants of 53. doi: blood pressure and other cardiovascular risk 10.1161/CIRCULATIONAHA.107.699579. factors (GAPP). Swiss Med Wkly. PMID: 18212285. Exclusion Code: X3. 2013;143:w13728. doi: 87. Danzi G, Giordano A, Biondi-Zoccai G, et 10.4414/smw.2013.13728. PMID: al. [The DANAMI-3-PRIMULTI study]. 23299990. Exclusion Code: X3. Giornale italiano di cardiologia (2006); 80. Cook NR, Buring JE, Ridker PM. The effect 2016. p. 171-5. Exclusion Code: X1. of including C-reactive protein in 88. Darabont R, Tautu OF, Pop D, et al. Visit- cardiovascular risk prediction models for to-Visit Blood Pressure Variability and women. Ann Intern Med. 2006 Jul Independently Predict 4;145(1):21-9. PMID: 16818925. Exclusion Cardiovascular Risk Category in a General Code: X3. Population: Results from the SEPHAR II 81. Cooper JA, Miller GJ, Humphries SE. A Study. Hellenic J Cardiol. 2015 May- comparison of the PROCAM and Jun;56(3):208-16. PMID: 26021242. Framingham point-scoring systems for Exclusion Code: X3. estimation of individual risk of coronary 89. Daugherty SL, Magid DJ, Kikla JR, et al. heart disease in the Second Northwick Park Gender differences in the prognostic value Heart Study. Atherosclerosis. 2005 of exercise treadmill test characteristics. Am Jul;181(1):93-100. doi: Heart J. 2011 May;161(5):908-14. doi: 10.1016/j.atherosclerosis.2004.12.026. 10.1016/j.ahj.2011.01.021. PMID: PMID: 15939059. Exclusion Code: X3. 21570521. Exclusion Code: X2. 82. Courand PY, Jenck S, Bricca G, et al. R 90. Daviglus ML, Liao Y, Greenland P, et al. wave in aVL lead: an outstanding ECG Association of nonspecific minor ST-T index in hypertension. J Hypertens. 2014 abnormalities with cardiovascular mortality: Jun;32(6):1317-25. doi: the Chicago Western Electric Study. JAMA. 10.1097/hjh.0000000000000181. PMID: 1999 Feb 10;281(6):530-6. PMID: 24751594. Exclusion Code: X4. 10022109. Exclusion Code: X5. 83. Crow RS, Hannan PJ, Folsom AR. 91. De Bacquer D, De Backer G, Kornitzer M, Prognostic significance of corrected QT and et al. Prognostic value of ECG findings for corrected JT interval for incident coronary total, cardiovascular disease, and coronary heart disease in a general population sample heart disease death in men and women. stratified by presence or absence of wide

Screening for CVD Ri sk Wi th ECG 87 RTI–UNC EPC Appendix C. Excluded Studies

Heart. 1998 Dec;80(6):570-7. PMID: Framingham Heart Study. Hypertension. 10065025. Exclusion Code: X5. 2006 May;47(5):861-7. doi: 92. de Groot JR, Krul SP, Kroon S, et al. 10.1161/01.HYP.0000217141.20163.23. Unidentified candidates for cardiac PMID: 16585411. Exclusion Code: X5. resynchronization therapy: guideline 100. Dhoble A, Lahr BD, Allison TG, et al. adherence in a large academic outpatient Predicting long-term cardiovascular risk clinic in the Netherlands. Pacing Clin using the mayo clinic cardiovascular risk Electrophysiol. 2013 Jan;36(1):69-75. doi: score in a referral population. Am J Cardiol. 10.1111/pace.12021. PMID: 23078171. 2014 Sep 1;114(5):704-10. doi: Exclusion Code: X2. 10.1016/j.amjcard.2014.05.061. PMID: 93. de L, II, Verhagen HJ, Stolker RJ, et al. The 25052544. Exclusion Code: X2. value of treadmill exercise test parameters 101. Dhoble A, Lahr BD, Allison TG, et al. together in patients with known or suspected Cardiopulmonary fitness and heart rate peripheral arterial disease. Eur J Prev recovery as predictors of mortality in a Cardiol. 2012 Apr;19(2):192-8. doi: referral population. J Am Heart Assoc. 10.1177/1741826711399986. PMID: 2014;3(2):e000559. doi: 21450584. Exclusion Code: X2. 10.1161/jaha.113.000559. PMID: 24663334. 94. de Lemos JA, Ayers CR, Levine B, et al. Exclusion Code: X7. Multimodality Strategy for Cardiovascular 102. Diercks GF, Hillege HL, van Boven AJ, et Risk Assessment: Performance in 2 al. Microalbuminuria modifies the mortality Population-Based Cohorts. Circulation. risk associated with electrocardiographic 2017 May 30;135(22):2119-32. doi: ST-T segment changes. J Am Coll Cardiol. 10.1161/CIRCULATIONAHA.117.027272. 2002 Oct 16;40(8):1401. PMID: 12392828. PMID: 28360032. Exclusion Code: X4. Exclusion Code: X5. 95. de Ruijter W, Westendorp RG, Assendelft 103. Donnan PT, Donnelly L, New JP, et al. WJ, et al. Use of Framingham risk score and Derivation and validation of a prediction new biomarkers to predict cardiovascular score for major coronary heart disease mortality in older people: population based events in a U.K. type 2 diabetic population. observational cohort study. BMJ. 2009 Jan Diabetes Care. 2006 Jun;29(6):1231-6. doi: 08;338:a3083. doi: 10.1136/bmj.a3083. 10.2337/dc05-1911. PMID: 16732001. PMID: 19131384. Exclusion Code: X3. Exclusion Code: X3. 96. DeFina L, Radford N, Leonard D, et al. 104. Ducloux D, Kazory A, Chalopin JM. Cardiorespiratory fitness and coronary artery Predicting coronary heart disease in renal calcification in women. Atherosclerosis. transplant recipients: a prospective study. 2014 Apr;233(2):648-53. doi: Kidney Int. 2004 Jul;66(1):441-7. doi: 10.1016/j.atherosclerosis.2014.01.016. 10.1111/j.1523-1755.2004.00751.x. PMID: PMID: 24561492. Exclusion Code: X7. 15200454. Exclusion Code: X2. 97. Denes P, Garside DB, Lloyd-Jones D, et al. 105. Dunder K, Lind L, Zethelius B, et al. Major and minor electrocardiographic Evaluation of a scoring scheme, including abnormalities and their association with proinsulin and the apolipoprotein underlying cardiovascular disease and risk B/ ratio, for the risk of factors in Hispanics/Latinos (from the acute coronary events in middle-aged men: Hispanic Community Health Study/Study of Uppsala Longitudinal Study of Adult Men Latinos). Am J Cardiol. 2013 Nov (ULSAM). Am Heart J. 2004 15;112(10):1667-75. doi: Oct;148(4):596-601. doi: 10.1016/j.amjcard.2013.08.004. PMID: 10.1016/j.ahj.2004.03.021. PMID: 24055066. Exclusion Code: X4. 15459588. Exclusion Code: X3. 98. Detrano RC, Wong ND, Doherty TM, et al. 106. Dyakova M, Shantikumar S, Colquitt Jill L, Coronary calcium does not accurately et al. Systematic versus opportunistic risk predict near-term future coronary events in assessment for the primary prevention of high-risk adults. Circulation. 1999 May cardiovascular disease. Cochrane Database 25;99(20):2633-8. PMID: 10338455. of Systematic Reviews. 2016(1)doi: Exclusion Code: X3. 10.1002/14651858.CD010411.pub2. PMID: 99. Dhingra R, Pencina MJ, Wang TJ, et al. CD010411. Exclusion Code: X7. Electrocardiographic QRS duration and the 107. Ekelund LG, Suchindran CM, McMahon risk of congestive heart failure: the RP, et al. Coronary heart disease morbidity

Screening for CVD Ri sk Wi th ECG 88 RTI–UNC EPC Appendix C. Excluded Studies

and mortality in hypercholesterolemic men associated with a high glomerular filtration predicted from an exercise test: the Lipid rate in the nondiabetic general population. Research Clinics Coronary Primary Kidney Int. 2014 Jul;86(1):146-53. doi: Prevention Trial. J Am Coll Cardiol. 1989 10.1038/ki.2013.470. PMID: 24304885. Sep;14(3):556-63. PMID: 2768706. Exclusion Code: X5. Exclusion Code: X4. 116. Erikssen G, Liestol K, Gullestad L, et al. 108. Elkeles RS, Godsland IF, Feher MD, et al. The terminal part of the QT interval (T peak Coronary calcium measurement improves to T end): a predictor of mortality after acute prediction of cardiovascular events in myocardial infarction. Ann Noninvasive asymptomatic patients with type 2 diabetes: Electrocardiol. 2012 Apr;17(2):85-94. doi: the PREDICT study. Eur Heart J. 2008 10.1111/j.1542-474X.2012.00493.x. PMID: Sep;29(18):2244-51. doi: 22537325. Exclusion Code: X2. 10.1093/eurheartj/ehn279. PMID: 117. Estes EH, Zhang ZM, Li Y, et al. The 18573867. Exclusion Code: X3. Romhilt-Estes left ventricular hypertrophy 109. Elsharawy MA, Al-Elq AH, Alkhadra AH, score and its components predict all-cause et al. Screening for asymptomatic mortality in the general population. Am cardiovascular disease in Arab patients with Heart J. 2015 Jul;170(1):104-9. doi: diabetes. Int Angiol. 2011 Feb;30(1):52-7. 10.1016/j.ahj.2015.04.004. PMID: PMID: 21248673. Exclusion Code: X4. 26093870. Exclusion Code: X5. 110. Empana JP, Ducimetiere P, Arveiler D, et al. 118. Fagher K, Nilsson A, Londahl M. Heart Are the Framingham and PROCAM rate-corrected QT interval prolongation as a coronary heart disease risk functions prognostic marker for 3-year survival in applicable to different European people with Type 2 diabetes undergoing populations? The PRIME Study. Eur Heart above-ankle amputation. Diabet Med. 2015 J. 2003 Nov;24(21):1903-11. PMID: May;32(5):679-85. doi: 10.1111/dme.12632. 14585248. Exclusion Code: X3. PMID: 25388827. Exclusion Code: X5. 111. Engeseth K, Hodnesdal C, Grundvold I, et 119. Faramawi MF, Caffrey JL, Amanzadeh J, et al. Heart rate reserve predicts cardiovascular al. estimated renal dysfunction death among physically unfit but otherwise predicts T wave axis deviation in US adults: healthy middle-aged men: a 35-year follow- results from NHANES III. Eur J Epidemiol. up study. Eur J Prev Cardiol. 2016 2011 Feb;26(2):101-7. doi: 10.1007/s10654- Jan;23(1):59-66. doi: 010-9534-5. PMID: 21184143. Exclusion 10.1177/2047487314553202. PMID: Code: X2. 25281482. Exclusion Code: X5. 120. Farrell SW, Finley CE, Haskell WL, et al. Is 112. Eranti A, Aro AL, Kerola T, et al. There a Gradient of Mortality Risk among Prevalence and prognostic significance of Men with Low Cardiorespiratory Fitness? abnormal P terminal force in lead V1 of the Med Sci Sports Exerc. 2015 Sep;47(9):1825- ECG in the general population. Circ 32. doi: 10.1249/mss.0000000000000608. Arrhythm Electrophysiol. 2014 PMID: 25551401. Exclusion Code: X5. Dec;7(6):1116-21. doi: 121. Farrell SW, Finley CE, McAuley PA, et al. 10.1161/circep.114.001557. PMID: Cardiorespiratory fitness, different measures 25381332. Exclusion Code: X4. of adiposity, and total cancer mortality in 113. Erdem A, Uenishi M, Kucukdurmaz Z, et al. women. Obesity (Silver Spring). 2011 The effect of metabolic syndrome on heart Nov;19(11):2261-7. doi: rate turbulence in non-diabetic patients. 10.1038/oby.2010.345. PMID: 21293448. Cardiol J. 2012;19(5):507-12. PMID: Exclusion Code: X5. 23042315. Exclusion Code: X5. 122. Farrell SW, Finley CE, Radford NB, et al. 114. Erez A, Kivity S, Berkovitch A, et al. The Cardiorespiratory fitness, body mass index, association between cardiorespiratory fitness and heart failure mortality in men: Cooper and cardiovascular risk may be modulated Center Longitudinal Study. Circ Heart Fail. by known cardiovascular risk factors. Am 2013 Sep 1;6(5):898-905. doi: Heart J. 2015 Jun;169(6):916-23.e1. doi: 10.1161/circheartfailure.112.000088. PMID: 10.1016/j.ahj.2015.02.023. PMID: 23873472. Exclusion Code: X4. 26027631. Exclusion Code: X5. 123. Fatemi O, Yuriditsky E, Tsioufis C, et al. 115. Eriksen BO, Lochen ML, Arntzen KA, et al. Impact of intensive glycemic control on the Subclinical cardiovascular disease is incidence of atrial fibrillation and associated

Screening for CVD Ri sk Wi th ECG 89 RTI–UNC EPC Appendix C. Excluded Studies

cardiovascular outcomes in patients with manifestations among Texas blood donors. type 2 diabetes mellitus (from the Action to Am J Cardiol. 2015 Jan 1;115(1):113-7. doi: Control Cardiovascular Risk in Diabetes 10.1016/j.amjcard.2014.09.050. PMID: Study). Am J Cardiol. 2014 Oct 25456877. Exclusion Code: X7. 15;114(8):1217-22. doi: 132. Gaziano TA, Young CR, Fitzmaurice G, et 10.1016/j.amjcard.2014.07.045. PMID: al. Laboratory-based versus non-laboratory- 25159234. Exclusion Code: X3. based method for assessment of 124. Ferrario M, Chiodini P, Chambless LE, et al. cardiovascular disease risk: the NHANES I Prediction of coronary events in a low Follow-up Study cohort. Lancet. 2008 Mar incidence population. Assessing accuracy of 15;371(9616):923-31. doi: 10.1016/S0140- the CUORE Cohort Study prediction 6736(08)60418-3. PMID: 18342687. equation. Int J Epidemiol. 2005 Exclusion Code: X3. Apr;34(2):413-21. doi: 10.1093/ije/dyh405. 133. Gencer B, Butler J, Bauer DC, et al. PMID: 15659467. Exclusion Code: X3. Association of electrocardiogram 125. Ferrie JE, Singh-Manoux A, Kivimaki M, et abnormalities and incident heart failure al. Cardiorespiratory risk factors as events. Am Heart J. 2014 Jun;167(6):869- predictors of 40-year mortality in women 75.e3. doi: 10.1016/j.ahj.2014.03.020. and men. Heart. 2009 Aug;95(15):1250-7. PMID: 24890537. Exclusion Code: X4. doi: 10.1136/hrt.2008.164251. PMID: 134. Georgoulias P, Demakopoulos N, 19389720. Exclusion Code: X5. Valotassiou V, et al. Long-term prognostic 126. Fitzmaurice DA, McCahon D, Baker J, et al. value of heart-rate recovery after treadmill Is screening for AF worthwhile? Stroke risk testing in patients with diabetes mellitus. Int in a screened population from the SAFE J Cardiol. 2009 May 1;134(1):67-74. doi: study. Fam Pract. 2014 Jun;31(3):298-302. 10.1016/j.ijcard.2008.01.036. PMID: doi: 10.1093/fampra/cmu011. PMID: 18499284. Exclusion Code: X3. 24728774. Exclusion Code: X4. 135. Ghaffari S, Kazemi B, Aliakbarzadeh P. 127. Fleg JL, Gerstenblith G, Zonderman AB, et Abnormal heart rate recovery after exercise al. Prevalence and prognostic significance of predicts coronary artery disease severity. exercise-induced silent myocardial ischemia Cardiol J. 2011;18(1):47-54. PMID: detected by thallium scintigraphy and 21305485. Exclusion Code: X7. electrocardiography in asymptomatic 136. Giagnoni E, Secchi MB, Wu SC, et al. volunteers. Circulation. 1990 Prognostic value of exercise EKG testing in Feb;81(2):428-36. PMID: 2297853. asymptomatic normotensive subjects. A Exclusion Code: X4. prospective matched study. N Engl J Med. 128. Floyd JS, Sitlani CM, Wiggins KL, et al. 1983 Nov 3;309(18):1085-9. doi: Variation in resting heart rate over 4 years 10.1056/NEJM198311033091803. PMID: and the risks of myocardial infarction and 6621650. Exclusion Code: X5. death among older adults. Heart. 2015 137. Giustetto C, Drago S, Demarchi PG, et al. Jan;101(2):132-8. doi: 10.1136/heartjnl- Risk stratification of the patients with 2014-306046. PMID: 25214500. Exclusion Brugada type electrocardiogram: a Code: X5. community-based prospective study. 129. Fuchs T, Torjman A, Galitzkaya L, et al. Europace. 2009 Apr;11(4):507-13. doi: The clinical significance of ventricular 10.1093/europace/eup006. PMID: arrhythmias during an exercise test in non- 19193676. Exclusion Code: X2. competitive and competitive athletes. Isr 138. Gordon DJ, Ekelund LG, Karon JM, et al. Med Assoc J. 2011 Dec;13(12):735-9. Predictive value of the exercise tolerance PMID: 22332442. Exclusion Code: X5. test for mortality in North American men: 130. Fyfe-Johnson AL, Muller CJ, Alonso A, et the Lipid Research Clinics Mortality al. Heart Rate Variability and Incident Follow-up Study. Circulation. 1986 Stroke: The Atherosclerosis Risk in Aug;74(2):252-61. PMID: 3731417. Communities Study. Stroke. 2016 Exclusion Code: X5. Jun;47(6):1452-8. doi: 139. Gorodeski EZ, Ishwaran H, Blackstone EH, 10.1161/strokeaha.116.012662. PMID: et al. Quantitative electrocardiographic 27217501. Exclusion Code: X5. measures and long-term mortality in 131. Garcia MN, Murray KO, Hotez PJ, et al. exercise test patients with clinically normal Development of chagas cardiac resting electrocardiograms. Am Heart J.

Screening for CVD Ri sk Wi th ECG 90 RTI–UNC EPC Appendix C. Excluded Studies

2009 Jul;158(1):61-70.e1. doi: 14;163(10):1263-9. PMID: 11107461. 10.1016/j.ahj.2009.04.015. PMID: Exclusion Code: X3. 19540393. Exclusion Code: X2. 148. Grundy SM, Barlow CE, Farrell SW, et al. 140. Gottdiener JS, Arnold AM, Aurigemma GP, Cardiorespiratory fitness and metabolic risk. et al. Predictors of congestive heart failure in Am J Cardiol. 2012 Apr 1;109(7):988-93. the elderly: the Cardiovascular Health doi: 10.1016/j.amjcard.2011.11.031. PMID: Study. J Am Coll Cardiol. 2000 22221951. Exclusion Code: X4. May;35(6):1628-37. PMID: 10807470. 149. Gulati M, Arnsdorf MF, Shaw LJ, et al. Exclusion Code: X5. Prognostic value of the duke treadmill score 141. Grassi G, Cifkova R, Laurent S, et al. Blood in asymptomatic women. Am J Cardiol. pressure control and cardiovascular risk 2005 Aug 1;96(3):369-75. doi: profile in hypertensive patients from central 10.1016/j.amjcard.2005.03.078. PMID: and eastern European countries: results of 16054460. Exclusion Code: X4. the BP-CARE study. Eur Heart J. 2011 150. Gulati M, Black HR, Arnsdorf MF, et al. Jan;32(2):218-25. doi: Kidney dysfunction, cardiorespiratory 10.1093/eurheartj/ehq394. PMID: fitness, and the risk of death in women. J 21047877. Exclusion Code: X6. Womens Health (Larchmt). 2012 142. Greenland P, LaBree L, Azen SP, et al. Sep;21(9):917-24. doi: Coronary artery calcium score combined 10.1089/jwh.2011.3406. PMID: 22480201. with Framingham score for risk prediction in Exclusion Code: X4. asymptomatic individuals. JAMA. 2004 Jan 151. Gulati M, Pandey DK, Arnsdorf MF, et al. 14;291(2):210-5. doi: Exercise capacity and the risk of death in 10.1001/jama.291.2.210. PMID: 14722147. women: the St James Women Take Heart Exclusion Code: X3. Project. Circulation. 2003 Sep 143. Greenland P, Xie X, Liu K, et al. Impact of 30;108(13):1554-9. doi: minor electrocardiographic ST-segment 10.1161/01.CIR.0000091080.57509.E9. and/or T-wave abnormalities on PMID: 12975254. Exclusion Code: X5. cardiovascular mortality during long-term 152. Gulati M, Shaw LJ, Thisted RA, et al. Heart follow-up. Am J Cardiol. 2003 May rate response to exercise stress testing in 1;91(9):1068-74. PMID: 12714148. asymptomatic women: the st. James women Exclusion Code: X5. take heart project. Circulation. 2010 Jul 144. Greiser KH, Kluttig A, Schumann B, et al. 13;122(2):130-7. doi: Cardiovascular diseases, risk factors and 10.1161/circulationaha.110.939249. PMID: short-term heart rate variability in an elderly 20585008. Exclusion Code: X5. general population: the CARLA study 2002- 153. Gupta S, Rohatgi A, Ayers CR, et al. 2006. Eur J Epidemiol. 2009;24(3):123-42. Cardiorespiratory fitness and classification doi: 10.1007/s10654-009-9317-z. PMID: of risk of cardiovascular disease mortality. 19199053. Exclusion Code: X7. Circulation. 2011 Apr 5;123(13):1377-83. 145. Groepenhoff H, Vonk-Noordegraaf A, van doi: 10.1161/circulationaha.110.003236. de Veerdonk MC, et al. Prognostic relevance PMID: 21422392. Exclusion Code: X3. of changes in exercise test variables in 154. Gu zd er RN, Gat ling W, Mullee MA, et al. pulmonary arterial hypertension. PLoS One. Prognostic value of the Framingham 2013;8(9):e72013. doi: cardiovascular risk equation and the UKPDS 10.1371/journal.pone.0072013. PMID: risk engine for coronary heart disease in 24039732. Exclusion Code: X3. newly diagnosed Type 2 diabetes: results 146. Grossman C, Ehrlich S, Shemesh J, et al. from a United Kingdom study. Diabet Med. Coronary artery calcium and exercise 2005 May;22(5):554-62. doi: electrocardiogram as predictors of coronary 10.1111/j.1464-5491.2005.01494.x. PMID: events in asymptomatic adults. Am J 15842509. Exclusion Code: X3. Cardiol. 2015 Mar 15;115(6):745-50. doi: 155. Haataja P, Anttila I, Nikus K, et al. 10.1016/j.amjcard.2014.12.039. PMID: Prognostic implications of intraventricular 25616536. Exclusion Code: X4. conduction delays in a general population: 147. Grover SA, Dorais M, Paradis G, et al. Lipid the Health 2000 Survey. Ann Med. 2015 screening to prevent coronary artery disease: Feb;47(1):74-80. doi: a quantitative evaluation of evolving 10.3109/07853890.2014.985704. PMID: guidelines. CMAJ. 2000 Nov 25613171. Exclusion Code: X5.

Screening for CVD Ri sk Wi th ECG 91 RTI–UNC EPC Appendix C. Excluded Studies

156. Hadaegh F, Ehteshami-Afshar S, 164. Hillebrand S, Gast KB, de Mutsert R, et al. Hajebrahimi MA, et al. Silent coronary Heart rate variability and first cardiovascular artery disease and incidence of event in populations without known cardiovascular and mortality events at cardiovascular disease: meta-analysis and different levels of glucose regulation; results dose-response meta-regression. Europace. of greater than a decade follow-up. Int J 2013 May;15(5):742-9. doi: Cardiol. 2015 Mar 1;182:334-9. doi: 10.1093/europace/eus341. PMID: 10.1016/j.ijcard.2015.01.017. PMID: 23370966. Exclusion Code: X7. 25585379. Exclusion Code: X5. 165. Hippisley-Cox J, Coupland C, Vinogradova 157. Hadaegh F, Hatami M, Mohebi R, et al. Y, et al. Performance of the QRISK Electrocardiography-defined silent CHD and cardiovascular risk prediction algorithm in risk of cardiovascular events among diabetic an independent UK sample of patients from patients in a Middle Eastern population. Eur general practice: a validation study. Heart. J Prev Cardiol. 2012 Dec;19(6):1227-33. 2008 Jan;94(1):34-9. doi: doi: 10.1177/1741826711428065. PMID: 10.1136/hrt.2007.134890. PMID: 17916661. 22013153. Exclusion Code: X2. Exclusion Code: X3. 158. Hage FG, Lusa L, Dondi M, et al. Exercise 166. Hippisley-Cox J, Coupland C, Vinogradova stress tests for detecting myocardial Y, et al. Derivation and validation of ischemia in asymptomatic patients with QRISK, a new cardiovascular disease risk diabetes mellitus. Am J Cardiol. 2013 Jul score for the United Kingdom: prospective 1;112(1):14-20. doi: open cohort study. BMJ. 2007 Jul 10.1016/j.amjcard.2013.02.047. PMID: 21;335(7611):136. doi: 23578350. Exclusion Code: X4. 10.1136/bmj.39261.471806.55. PMID: 159. Halon DA, Dobrecky-Mery I, Gaspar T, et 17615182. Exclusion Code: X3. al. Heart rate recovery after exercise and 167. Hippisley-Cox J, Coupland C, Vinogradova coronary in asymptomatic Y, et al. Predicting cardiovascular risk in individuals with type 2 diabetes mellitus: a England and Wales: prospective derivation study using 64-slice coronary CT and validation of QRISK2. BMJ. 2008 Jun angiography. Int J Cardiol. 2010 Nov 28;336(7659):1475-82. doi: 5;145(1):102-3. doi: 10.1136/bmj.39609.449676.25. PMID: 10.1016/j.ijcard.2009.05.045. PMID: 18573856. Exclusion Code: X3. 19540605. Exclusion Code: X5. 168. Hisamatsu T, Miura K, Ohkubo T, et al. 160. Hankil Kim J, Dooley P, Smith R. Clinical Interaction between dietary marine-derived Inquiry: Do asymptomatic adults need n-3 fatty acids intake and J-point elevation screening EKGs? J Fam Pract. 2013 on the risk of cardiac death: a 24-year Aug;62(8):438-50. PMID: 24143339. follow-up of Japanese men. Heart. 2013 Exclusion Code: X7. Jul;99(14):1024-9. doi: 10.1136/heartjnl- 161. Hartaigh B, Allore HG, Trentalange M, et 2012-303496. PMID: 23666393. Exclusion al. Elevations in time-varying resting heart Code: X4. rate predict subsequent all-cause mortality in 169. Hisamatsu T, Ohkubo T, Miura K, et al. older adults. Eur J Prev Cardiol. 2015 Association between J-point elevation and Apr;22(4):527-34. doi: death from coronary artery disease--15-year 10.1177/2047487313519932. PMID: follow up of the NIPPON DATA90. Circ J. 24445263. Exclusion Code: X2. 2013;77(5):1260-6. PMID: 23358431. 162. Hense HW, Schulte H, Lowel H, et al. Exc lusion Code: X4. Framingham risk function overestimates risk 170. Ho JS, Fitzgerald SJ, Barlow CE, et al. Risk of coronary heart disease in men and women of mortality increases with increasing from Germany--results from the MONICA number of abnormal non-ST parameters Augsburg and the PROCAM cohorts. Eur recorded during exercise testing. Eur J Heart J. 2003 May;24(10):937-45. PMID: Cardiovasc Prev Rehabil. 2010 12714025. Exclusion Code: X3. Aug;17(4):462-8. doi: 163. Heston TF. Standardizing predictive values 10.1097/HJR.0b013e328336a10d. PMID: in diagnostic imaging research. J Magn 20084008. Exclusion Code: X5. Reson Imaging. 2011 Feb;33(2):505; author 171. Hodnesdal C, Prestgaard E, Erikssen G, et reply 6-7. doi: 10.1002/jmri.22466. PMID: al. Rapidly upsloping ST-segment on 21274995. Exclusion Code: X7. exercise ECG: a marker of reduced coronary

Screening for CVD Ri sk Wi th ECG 92 RTI–UNC EPC Appendix C. Excluded Studies

heart disease mortality risk. Eur J Prev 2012 Nov;14(11):1285-92. doi: Cardiol. 2013 Aug;20(4):541-8. doi: 10.1093/eurjhf/hfs112. PMID: 22791081. 10.1177/2047487312444370. PMID: Exclusion Code: X5. 22492865. Exclusion Code: X4. 179. Ilkhanoff L, Soliman EZ, Prineas RJ, et al. 172. Hollenberg M, Zoltick JM, Go M, et al. Clinical characteristics and outcomes Comparison of a quantitative treadmill associated with the natural history of early exercise score with standard repolarization in a young, biracial cohort electrocardiographic criteria in screening followed to middle age: the Coronary Artery asymptomatic young men for coronary Risk Development in Young Adults artery disease. N Engl J Med. 1985 Sep (CARDIA) study. Circ Arrhythm 5;313(10):600-6. doi: Electrophysiol. 2014 Jun;7(3):392-9. doi: 10.1056/NEJM198509053131003. PMID: 10.1161/circep.113.000874. PMID: 4022047. Exclusion Code: X4. 24759868. Exclusion Code: X3. 173. Holtermann A, Mortensen OS, Burr H, et al. 180. Inama G, Pedrinazzi C. T-wave alternans Long work hours and physical fitness: 30- and ventricular arrhythmias in athletes: year risk of ischaemic heart disease and all- usefulness of follow-up updates. Ann cause mortality among middle-aged Noninvasive Electrocardiol. 2009 Caucasian men. Heart. 2010 Jan;14(1):99. doi: 10.1111/j.1542- Oct;96(20):1638-44. doi: 474X.2008.00280.x. PMID: 19149801. 10.1136/hrt.2010.197145. PMID: 20820054. Exclusion Code: X7. Exclusion Code: X3. 181. Inohara T, Kohsaka S, Okamura T, et al. 174. Holtermann A, Mortensen OS, Burr H, et al. Cumulative impact of axial, structural, and Physical work demands and physical fitness repolarization ECG findings on long-term in low social classes--30-year ischemic heart cardiovascular mortality among healthy disease and all-cause mortality in the individuals in Japan: National Integrated Copenhagen Male Study. J Occup Environ Project for Prospective Observation of Non- Med. 2011 Nov;53(11):1221-7. doi: Communicable Disease and its Trends in the 10.1097/JOM.0b013e318233865f. PMID: Aged, 1980 and 1990. Eur J Prev Cardiol. 22015549. Exclusion Code: X3. 2014 Dec;21(12):1501-8. doi: 175. Hung RK, Al-Mallah MH, McEvoy JW, et 10.1177/2047487313500568. PMID: al. Prognostic value of exercise capacity in 23918839. Exclusion Code: X4. patients with coronary artery disease: the 182. Inohara T, Kohsaka S, Okamura T, et al. FIT (Henry Ford ExercIse Testing) project. Long-term outcome of healthy participants Mayo Clin Proc. 2014 Dec;89(12):1644-54. with atrial premature complex: a 15-year doi: 10.1016/j.mayocp.2014.07.011. PMID: follow-up of the NIPPON DATA 90 cohort. 25440889. Exclusion Code: X2. PLoS One. 2013;8(11):e80853. doi: 176. Hung RK, Al-Mallah MH, Qadi MA, et al. 10.1371/journal.pone.0080853. PMID: Cardiorespiratory fitness attenuates risk for 24260495. Exclusion Code: X4. major adverse cardiac events in 183. Iribarren C, Round AD, Peng JA, et al. Short hyperlipidemic men and women QT in a cohort of 1.7 million persons: independent of statin therapy: The Henry prevalence, correlates, and prognosis. Ann Ford ExercIse Testing Project. Am Heart J. Noninvasive Electrocardiol. 2014 2015 Aug;170(2):390-9. doi: Sep;19(5):490-500. doi: 10.1016/j.ahj.2015.04.030. PMID: 10.1111/anec.12157. PMID: 24829126. 26299238. Exclusion Code: X2. Exclusion Code: X7. 177. Ilias NA, Xian H, Inman C, et al. Arm 184. Ishikawa J, Ishikawa S, Kario K. Levels of exercise testing predicts clinical outcome. cornell voltage and cornell product for Am Heart J. 2009 Jan;157(1):69-76. doi: predicting cardiovascular and stroke 10.1016/j.ahj.2008.09.007. PMID: mortality and morbidity in the general 19081399. Exclusion Code: X2. Japanese population. Circ J. 178. Ilkhanoff L, Liu K, Ning H, et al. 2014;78(2):465-75. PMID: 24284883. Association of QRS duration with left Exclusion Code: X4. ventricular structure and function and risk of 185. Jacob MJ. Evaluation of TMT abnormalities heart failure in middle-aged and older in asymptomatic persons using myocardial adults: the Multi-Ethnic Study of perfusion study. J Assoc Physicians India. Atherosclerosis (MESA). Eur J Heart Fail.

Screening for CVD Ri sk Wi th ECG 93 RTI–UNC EPC Appendix C. Excluded Studies

2011 Mar;59:155-6, 61-3. PMID: 21751624. 10.1056/NEJMoa043012. PMID: 15888695. Exclusion Code: X5. Exclusion Code: X5. 186. Jee SH, Park JW, Lee SY, et al. Stroke risk 194. Jouven X, Zureik M, Desnos M, et al. Long- prediction model: a risk profile from the term outcome in asymptomatic men with Korean study. Atherosclerosis. 2008 exercise-induced premature ventricular Mar;197(1):318-25. doi: depolarizations. N Engl J Med. 2000 Sep 10.1016/j.atherosclerosis.2007.05.014. 21;343(12):826-33. doi: PMID: 17586511. Exclusion Code: X3. 10.1056/NEJM200009213431201. PMID: 187. Jensen MT, Suadicani P, Hein HO, et al. 10995861. Exclusion Code: X4. Elevated resting heart rate, physical fitness 195. Julius S, Palatini P, Kjeldsen SE, et al. and all-cause mortality: a 16-year follow-up Usefulness of heart rate to predict cardiac in the Copenhagen Male Study. Heart. 2013 events in treated patients with high-risk Jun;99(12):882-7. doi: 10.1136/heartjnl- systemic hypertension. Am J Cardiol. 2012 2012-303375. PMID: 23595657. Exclusion Mar 1;109(5):685-92. doi: Code: X4. 10.1016/j.amjcard.2011.10.025. PMID: 188. Jeong HC, Kim I, Park KH, et al. New 22169130. Exclusion Code: X2. strategy for detection of subclinical coronary 196. Junttila MJ, Tikkanen JT, Kentta T, et al. atherosclerosis in asymptomatic patients Early repolarization as a predictor of with type 2 diabetes based on cardiac multi- arrhythmic and nonarrhythmic cardiac detector computed tomography and events in middle-aged subjects. Heart treadmill test. Circ J. 2014;78(3):671-8. Rhythm. 2014 Oct;11(10):1701-6. doi: PMID: 24401570. Exclusion Code: X4. 10.1016/j.hrthm.2014.05.024. PMID: 189. Jimenez-Pavon D, Artero EG, Lee DC, et al. 24858812. Exclusion Code: X5. Cardiorespiratory Fitness and Risk of 197. Kabutoya T, Ishikawa S, Ishikawa J, et al. P- Sudden Cardiac Death in Men and Women wave morphologic characteristics predict in the United States: A Prospective cardiovascular events in a community- Evaluation From the Aerobics Center dwelling population. Ann Noninvasive Longitudinal Study. Mayo Clin Proc. 2016 Electrocardiol. 2012 Jul;17(3):252-9. doi: Jul;91(7):849-57. doi: 10.1111/j.1542-474X.2012.00529.x. PMID: 10.1016/j.mayocp.2016.04.025. PMID: 22816544. Exclusion Code: X7. 27378037. Exclusion Code: X5. 198. Kahn S, Frishman WH, Weissman S, et al. 190. Jissho S, Shimada K, Taguchi H, et al. Left ventricular hypertrophy on Impact of electrocardiographic left electrocardiogram: prognostic implications ventricular hypertrophy on the occurrence of from a 10-year cohort study of older cardiovascular events in elderly subjects: a report from the Bronx hypertensive patients. - The Japanese trial to Longitudinal Aging Study. J Am Geriatr assess optimal systolic blood pressure in Soc. 1996 May;44(5):524-9. PMID: elderly hypertensive patients (JATOS). Circ 8617900. Exclusion Code: X5. J. 2010 May;74(5):938-45. PMID: 199. Kamel H, Bartz TM, Longstreth WT, Jr., et 20339195. Exclusion Code: X2. al. Association between left atrial 191. Johnson NP, Goldberger JJ. Prognostic abnormality on ECG and vascular brain value of late heart rate recovery after injury on MRI in the Cardiovascular Health treadmill exercise. Am J Cardiol. 2012 Jul Study. Stroke. 2015 Mar;46(3):711-6. doi: 1;110(1):45-9. doi: 10.1161/strokeaha.114.007762. PMID: 10.1016/j.amjcard.2012.02.046. PMID: 25677594. Exclusion Code: X5. 22463837. Exclusion Code: X2. 200. Kamel H, Hunter M, Moon YP, et al. 192. Josephson RA, Shefrin E, Lakatta EG, et al. Electrocardiographic Left Atrial Can serial exercise testing improve the Abnormality and Risk of Stroke: Northern prediction of coronary events in Manhattan Study. Stroke. 2015 asymptomatic individuals? Circulation. Nov;46(11):3208-12. doi: 1990 Jan;81(1):20-4. PMID: 2297826. 10.1161/strokeaha.115.009989. PMID: Exclusion Code: X4. 26396031. Exclusion Code: X5. 193. Jouven X, Empana JP, Schwartz PJ, et al. 201. Kamel H, O'Neal WT, Okin PM, et al. Heart-rate profile during exercise as a Electrocardiographic left atrial abnormality predictor of sudden death. N Engl J Med. and stroke subtype in the atherosclerosis risk 2005 May 12;352(19):1951-8. doi: in communities study. Ann Neurol. 2015

Screening for CVD Ri sk Wi th ECG 94 RTI–UNC EPC Appendix C. Excluded Studies

Nov;78(5):670-8. doi: 10.1002/ana.24482. 210. Khan HS, Iftikhar I, Khan Q. Validity of PMID: 26179566. Exclusion Code: X5. Electrocardiographic QT Interval in 202. Kamel H, Soliman EZ, Heckbert SR, et al. Predicting Left Ventricular Diastolic P-wave morphology and the risk of incident Dysfunction in Patients with Suspected ischemic stroke in the Multi-Ethnic Study of Heart Failure. J Coll Physicians Surg Pak. Atherosclerosis. Stroke. 2014 2016 May;26(5):353-6. doi: 2313. PMID: Sep;45(9):2786-8. doi: 27225136. Exclusion Code: X2. 10.1161/strokeaha.114.006364. PMID: 211. Khan R, Jang IK. Evaluation of coronary 25052322. Exclusion Code: X5. allograft vasculopathy using multi-detector 203. Kanaya AM, Kandula N, Herrington D, et row computed tomography: a systematic al. Mediators of Atherosclerosis in South review (Provisional abstract). European Asians Living in America (MASALA) Journal of Cardio-Thoracic Surgery. study: objectives, methods, and cohort 2012;41(2):415-22. PMID: DARE- description. Clin Cardiol. 2013 12012022142. Exclusion Code: X2. Dec;36(12):713-20. doi: 10.1002/clc.22219. 212. Kim JH, Baggish AL. Electrocardiographic PMID: 24194499. Exclusion Code: X5. right and patterns 204. Katsikis A, Theodorakos A, Kouzoumi A, et in athletes: prevalence, pathology, and al. Prognostic value of the Duke treadmill clinical significance. J Electrocardiol. 2015 score in octogenarians undergoing May-Jun;48(3):380-4. doi: myocardial perfusion imaging. 10.1016/j.jelectrocard.2015.03.015. PMID: Atherosclerosis. 2014 Oct;236(2):373-80. 25836379. Exclusion Code: X7. doi: 10.1016/j.atherosclerosis.2014.07.028. 213. Kobal SL, Wilkof-Segev R, Patchett MS, et PMID: 25133351. Exclusion Code: X2. al. Prognostic value of myocardial ischemic 205. Katsikis A, Theodorakos A, Papaioannou S, electrocardiographic response in patients et al. Long-term prognostic value of with normal stress echocardiographic study. myocardial perfusion imaging in Am J Cardiol. 2014 Mar 15;113(6):945-9. octogenarians able to undergo treadmill doi: 10.1016/j.amjcard.2013.11.051. PMID: exercise stress testing. J Nucl Cardiol. 2014 24440328. Exclusion Code: X2. Dec;21(6):1213-22. doi: 10.1007/s12350- 214. Koenig W, Lowel H, Baumert J, et al. C- 014-9991-4. PMID: 25189145. Exclusion reactive protein modulates risk prediction Code: X2. based on the Framingham Score: 206. Kawasaki-Ogita Y, Hamamoto Y, Honjo S, implications for future risk assessment: et al. The limited usefulness of the treadmill results from a large cohort study in southern test or a risk-guided approach in screening Germany. Circulation. 2004 Mar for asymptomatic coronary heart disease in 23;109(11):1349-53. doi: Japanese patients with type 2 diabetes. 10.1161/01.CIR.0000120707.98922.E3. Intern Med. 2012;51(24):3337-42. PMID: PMID: 15023871. Exclusion Code: X3. 23257517. Exclusion Code: X5. 215. Kokkinos P, Doumas M, Myers J, et al. A 207. Kellett J, Rasool S, McLoughlin B. graded association of exercise capacity and Prediction of mortality 1 year after hospital all-cause mortality in males with high- admission. Qjm. 2012 Sep;105(9):847-53. normal blood pressure. Blood Press. doi: 10.1093/qjmed/hcs099. PMID: 2009;18(5):261-7. doi: 22690010. Exclusion Code: X2. 10.3109/08037050903272859. PMID: 208. Kentta T, Karsikas M, Junttila MJ, et al. 19919397. Exclusion Code: X4. QRS-T morphology measured from exercise 216. Kokkinos P, Faselis C, Myers J, et al. Statin electrocardiogram as a predictor of cardiac therapy, fitness, and mortality risk in mortality. Europace. 2011 May;13(5):701-7. middle-aged hypertensive male veterans. Am doi: 10.1093/europace/euq461. PMID: J Hypertens. 2014 Mar;27(3):422-30. doi: 21186225. Exclusion Code: X5. 10.1093/ajh/hpt241. PMID: 24436326. 209. Khan H, Kunutsor S, Rauramaa R, et al. Exclusion Code: X2. Cardiorespiratory fitness and risk of heart 217. Kokkinos P, Myers J, Doumas M, et al. failure: a population-based follow-up study. Heart rate recovery, exercise capacity, and Eur J Heart Fail. 2014 Feb;16(2):180-8. mortality risk in male veterans. Eur J Prev doi: 10.1111/ejhf.37. PMID: 24464981. Cardiol. 2012 Apr;19(2):177-84. doi: Exclusion Code: X2. 10.1177/1741826711398432. PMID: 21450594. Exclusion Code: X4.

Screening for CVD Ri sk Wi th ECG 95 RTI–UNC EPC Appendix C. Excluded Studies

218. Kokkinos PF, Faselis C, Myers J, et al. 2015 Feb;47(1):69-73. doi: Interactive effects of fitness and statin 10.3109/07853890.2014.985703. PMID: treatment on mortality risk in veterans with 25613172. Exclusion Code: X5. dyslipidaemia: a cohort study. Lancet. 2013 227. Kurl S, Makikallio TH, Rautaharju P, et al. Feb 2;381(9864):394-9. doi: 10.1016/s0140- Duration of QRS complex in resting 6736(12)61426-3. PMID: 23199849. electrocardiogram is a predictor of sudden Exclusion Code: X2. cardiac death in men. Circulation. 2012 May 219. Koller MT, Steyerberg EW, Wolbers M, et 29;125(21):2588-94. doi: al. Validity of the Framingham point scores 10.1161/circulationaha.111.025577. PMID: in the elderly: results from the Rotterdam 22615341. Exclusion Code: X2. study. Am Heart J. 2007 Jul;154(1):87-93. 228. Kurl S, Sivenius J, Makikallio TH, et al. doi: 10.1016/j.ahj.2007.03.022. PMID: Exercise workload, cardiovascular risk 17584559. Exclusion Code: X3. factor evaluation and the risk of stroke in 220. Kook HY, Jeong MH, Oh S, et al. Current middle-aged men. J Intern Med. 2009 trend of acute myocardial infarction in Feb;265(2):229-37. doi: 10.1111/j.1365- Korea (from the Korea Acute Myocardial 2796.2008.02006.x. PMID: 18793247. Infarction Registry from 2006 to 2013). Am Exclusion Code: X5. J Cardiol. 2014 Dec 15;114(12):1817-22. 229. Ladapo JA, Blecker S, Douglas PS. doi: 10.1016/j.amjcard.2014.09.019. PMID: Physician decision making and trends in the 25438907. Exclusion Code: X2. use of cardiac stress testing in the United 221. Kop WJ, Stein PK, Tracy RP, et al. States: an analysis of repeated cross- Autonomic nervous system dysfunction and sectional data. Ann Intern Med. 2014 Oct inflammation contribute to the increased 7;161(7):482-90. doi: 10.7326/m14-0296. cardiovascular mortality risk associated with PMID: 25285541. Exclusion Code: X7. depression. Psychosom Med. 2010 230. Ladapo JA, Blecker S, Elashoff MR, et al. Sep;72(7):626-35. doi: Clinical implications of referral bias in the 10.1097/PSY.0b013e3181eadd2b. PMID: diagnostic performance of exercise testing 20639389. Exclusion Code: X3. for coronary artery disease (Provisional 222. Kothari V, Stevens RJ, Adler AI, et al. abstract). Database of Abstracts of Reviews UKPDS 60: risk of stroke in type 2 diabetes of Effects. 2013(2):e000505. PMID: DARE- estimated by the UK Prospective Diabetes 12014040303. Exclusion Code: X3. Study risk engine. Stroke. 2002 231. Lakoski SG, Barlow CE, Farrell SW, et al. Jul;33(7):1776-81. PMID: 12105351. Impact of body mass index, physical Exclusion Code: X3. activity, and other clinical factors on 223. Krepp JM, Lin F, Min JK, et al. Relationship cardiorespiratory fitness (from the Cooper of electrocardiographic left ventricular Center longitudinal study). Am J Cardiol. hypertrophy to the presence of diastolic 2011 Jul 1;108(1):34-9. doi: dysfunction. Ann Noninvasive 10.1016/j.amjcard.2011.02.338. PMID: Electrocardiol. 2014 Nov;19(6):552-60. doi: 21529738. Exclusion Code: X5. 10.1111/anec.12166. PMID: 24750238. 232. Lakoski SG, Willis BL, Barlow CE, et al. Exclusion Code: X4. Midlife Cardiorespiratory Fitness, Incident 224. Krijthe BP, Leening MJ, Heeringa J, et al. Cancer, and Survival After Cancer in Men: Unrecognized myocardial infarction and risk The Cooper Center Longitudinal Study. of atrial fibrillation: the Rotterdam Study. JAMA Oncol. 2015 May;1(2):231-7. doi: Int J Cardiol. 2013 Sep 30;168(2):1453-7. 10.1001/jamaoncol.2015.0226. PMID: doi: 10.1016/j.ijcard.2012.12.057. PMID: 26181028. Exclusion Code: X5. 23332895. Exclusion Code: X5. 233. Larsen BS, Kumarathurai P, Falkenberg J, et 225. Kurl S, Laukkanen JA, Tuomainen TP, et al. al. Excessive Atrial Ectopy and Short Atrial Association of exercise-induced, silent ST- Runs Increase the Risk of Stroke Beyond segment depression with the risk of stroke Incident Atrial Fibrillation. J Am Coll and cardiovascular diseases in men. Stroke. Cardiol. 2015 Jul 21;66(3):232-41. doi: 2003 Jul;34(7):1760-5. doi: 10.1016/j.jacc.2015.05.018. PMID: 10.1161/01.STR.0000078564.46376.0A. 26184616. Exclusion Code: X5. PMID: 12829872. Exclusion Code: X5. 234. Larsen CT, Dahlin J, Blackburn H, et al. 226. Kurl S, Makikallio TH, Laukkanen JA. T- Prevalence and prognosis of wave inversion and mortality risk. Ann Med. electrocardiographic left ventricular

Screening for CVD Ri sk Wi th ECG 96 RTI–UNC EPC Appendix C. Excluded Studies

hypertrophy, ST segment depression and 242. Lee CD, Sui X, Blair SN. Combined effects negative T-wave; the Copenhagen City of cardiorespiratory fitness, not smoking, Heart Study. Eur Heart J. 2002 and normal waist girth on morbidity and Feb;23(4):315-24. doi: mortality in men. Arch Intern Med. 2009 10.1053/euhj.2001.2774. PMID: 11812068. Dec 14;169(22):2096-101. doi: Exclusion Code: X5. 10.1001/archinternmed.2009.414. PMID: 235. Larstorp AC, Okin PM, Devereux RB, et al. 20008693. Exclusion Code: X5. Changes in electrocardiographic left 243. Lee DC, Sui X, Artero EG, et al. Long-term ventricular hypertrophy and risk of major effects of changes in cardiorespiratory cardiovascular events in isolated systolic fitness and body mass index on all-cause hypertension: the LIFE study. J Hum and cardiovascular disease mortality in men: Hypertens. 2011 Mar;25(3):178-85. doi: the Aerobics Center Longitudinal Study. 10.1038/jhh.2010.52. PMID: 20505749. Circulation. 2011 Dec 6;124(23):2483-90. Exclusion Code: X5. doi: 10.1161/circulationaha.111.038422. 236. Lauer MS, Okin PM, Larson MG, et al. PMID: 22144631. Exclusion Code: X3. Impaired heart rate response to graded 244. Lee DC, Sui X, Ortega FB, et al. exercise. Prognostic implications of Comparisons of leisure-time physical chronotropic incompetence in the activity and cardiorespiratory fitness as Framingham Heart Study. Circulation. 1996 predictors of all-cause mortality in men and Apr 15;93(8):1520-6. PMID: 8608620. women. Br J Sports Med. 2011 Exclusion Code: X5. May;45(6):504-10. doi: 237. Laukkanen JA, Di Angelantonio E, Khan H, 10.1136/bjsm.2009.066209. PMID: et al. T-wave inversion, QRS duration, and 20418526. Exclusion Code: X5. QRS/T angle as electrocardiographic 245. Lee ET, Howard BV, Wang W, et al. predictors of the risk for sudden cardiac Prediction of coronary heart disease in a death. Am J Cardiol. 2014 Apr population with high prevalence of diabetes 1;113(7):1178-83. doi: and albuminuria: the . 10.1016/j.amjcard.2013.12.026. PMID: Circulation. 2006 Jun 27;113(25):2897-905. 24513474. Exclusion Code: X2. doi: 238. Laukkanen JA, Kurl S, Lakka TA, et al. 10.1161/CIRCULATIONAHA.105.593178. Exercise-induced silent myocardial ischemia PMID: 16769914. Exclusion Code: X4. and coronary morbidity and mortality in 246. Lee JM, Yoo KD, Oh YS, et al. Relationship middle-aged men. J Am Coll Cardiol. 2001 between resting electrocardiographic Jul;38(1):72-9. PMID: 11451298. Exclusion parameters and estimated 10-year risk for Code: X5. coronary heart disease in healthy adults in 239. Laukkanen JA, Kurl S, Rauramaa R, et al. the USA. Ann Noninvasive Electrocardiol. Systolic blood pressure response to exercise 2010 Oct;15(4):315-20. doi: 10.1111/j.1542- testing is related to the risk of acute 474X.2010.00386.x. PMID: 20946553. myocardial infarction in middle-aged men. Exclusion Code: X5. Eur J Cardiovasc Prev Rehabil. 2006 247. Lee V, Hemingway H, Harb R, et al. The Jun;13(3):421-8. PMID: 16926673. prognostic significance of premature Exclusion Code: X5. ventricular complexes in adults without 240. Laukkanen JA, Rauramaa R, Kurl S. clinically apparent heart disease: a meta- Exercise workload, coronary risk evaluation analysis and systematic review. Heart. 2012 and the risk of cardiovascular and all-cause Sep;98(17):1290-8. doi: 10.1136/heartjnl- death in middle-aged men. Eur J Cardiovasc 2012-302005. PMID: 22781425. Exclusion Prev Rehabil. 2008 Jun;15(3):285-92. doi: Code: X7. 10.1097/HJR.0b013e3282f37a33. PMID: 248. Leigh JA, O'Neal WT, Soliman EZ. 18525382. Exclusion Code: X5. Electrocardiographic Left Ventricular 241. Laukkanen JA, Willeit P, Kurl S, et al. Hypertrophy as a Predictor of Elevated systolic blood pressure during Cardiovascular Disease Independent of Left recovery from exercise and the risk of Ventricular Anatomy in Subjects Aged sudden cardiac death. J Hypertens. 2014 >/=65 Years. Am J Cardiol. 2016 Jun Mar;32(3):659-66. doi: 01;117(11):1831-5. doi: 10.1097/hjh.0000000000000066. PMID: 10.1016/j.amjcard.2016.03.020. PMID: 24317550. Exclusion Code: X2. 27067620. Exclusion Code: X4.

Screening for CVD Ri sk Wi th ECG 97 RTI–UNC EPC Appendix C. Excluded Studies

249. Leo T, Uberoi A, Jain NA, et al. The impact 10.1097/HJR.0b013e328011490a. PMID: of ST elevation on athletic screening. Clin J 17667638. Exclusion Code: X3. Sport Med. 2011 Sep;21(5):433-40. doi: 258. Liu J, Hong Y, D'Agostino RB, Sr., et al. 10.1097/jsm.0b013e31822cf105. PMID: Predictive value for the Chinese population 21892017. Exclusion Code: X7. of the Framingham CHD risk assessment 250. Levitt K, Aves T, Dorian P, et al. Lack of tool compared with the Chinese Multi- difference in T wave variability between Provincial Cohort Study. JAMA. 2004 Jun patients at risk of sudden cardiac death and 2;291(21):2591-9. doi: healthy subjects. J Electrocardiol. 2014 10.1001/jama.291.21.2591. PMID: Mar-Apr;47(2):251-6. doi: 15173150. Exclusion Code: X3. 10.1016/j.jelectrocard.2013.12.007. PMID: 259. Liu R, Sui X, Laditka JN, et al. 24456792. Exclusion Code: X4. Cardiorespiratory fitness as a predictor of 251. Li Y, Dawood FZ, Chen H, et al. Minor dementia mortality in men and women. Med isolated Q waves and cardiovascular events Sci Sports Exerc. 2012 Feb;44(2):253-9. doi: in the MESA study. Am J Med. 2013 10.1249/MSS.0b013e31822cf717. PMID: May;126(5):450.e9-.e16. doi: 21796048. Exclusion Code: X5. 10.1016/j.amjmed.2012.10.030. PMID: 260. Lovett KM, Liang BA. Direct-to-consumer 23582938. Exclusion Code: X4. cardiac screening and suspect risk 252. Li Y, Shah AJ, Soliman EZ. Effect of evaluation. Jama. 2011 Jun electrocardiographic P-wave axis on 22;305(24):2567-8. doi: mortality. Am J Cardiol. 2014 Jan 10.1001/jama.2011.865. PMID: 21693746. 15;113(2):372-6. doi: Exclusion Code: X7. 10.1016/j.amjcard.2013.08.050. PMID: 261. Lowres N, Neubeck L, Redfern J, et al. 24176072. Exclusion Code: X2. Screening to identify unknown atrial 253. Liao Y, McGee DL, Cooper RS. Prediction fibrillation. A systematic review. Thromb of coronary heart disease mortality in blacks Haemost. 2013 Aug;110(2):213-22. doi: and whites: pooled data from two national 10.1160/th13-02-0165. PMID: 23595785. cohorts. Am J Cardiol. 1999 Jul 1;84(1):31- Exclusion Code: X7. 6. PMID: 10404847. Exclusion Code: X3. 262. Lyerly GW, Sui X, Church TS, et al. 254. Liao Y, McGee DL, Cooper RS, et al. How Maximal exercise electrocardiography generalizable are coronary risk prediction responses and coronary heart disease models? Comparison of Framingham and mortality among men with diabetes mellitus. two national cohorts. Am Heart J. 1999 Circulation. 2008 May 27;117(21):2734-42. May;137(5):837-45. PMID: 10220632. doi: Exclusion Code: X3. 10.1161/CIRCULATIONAHA.107.729277. 255. Liao YL, Liu KA, Dyer A, et al. Major and PMID: 18490521. Exclusion Code: X5. minor electrocardiographic abnormalities 263. Lyerly GW, Sui X, Church TS, et al. and risk of death from coronary heart Maximal exercise electrocardiographic disease, cardiovascular diseases and all responses and coronary heart disease causes in men and women. J Am Coll mortality among men with metabolic Cardiol. 1988 Dec;12(6):1494-500. PMID: syndrome. Mayo Clin Proc. 2010 3192848. Exclusion Code: X5. Mar;85(3):239-46. doi: 256. Lindekleiv H, Wilsgaard T, Macfarlane PW, 10.4065/mcp.2009.0509. PMID: 20160139. et al. QT interval and the risk of myocardial Exclusion Code: X5. infarction and all-cause death: a cohort 264. Lyerly GW, Sui X, Lavie CJ, et al. The study. J Cardiovasc Electrophysiol. 2012 association between cardiorespiratory fitness Aug;23(8):846-52. doi: 10.1111/j.1540- and risk of all-cause mortality among 8167.2012.02308.x. PMID: 22509793. women with impaired fasting glucose or Exclusion Code: X4. undiagnosed diabetes mellitus. Mayo Clin 257. Lindman AS, Veierod MB, Pedersen JI, et Proc. 2009 Sep;84(9):780-6. doi: al. The ability of the SCORE high-risk 10.1016/s0025-6196(11)60487-4. PMID: model to predict 10-year cardiovascular 19720775. Exclusion Code: X5. disease mortality in Norway. Eur J 265. Macfarlane PW, Murray H, Sattar N, et al. Cardiovasc Prev Rehabil. 2007 The incidence and risk factors for new onset Aug;14(4):501-7. doi: atrial fibrillation in the PROSPER study. Europace. 2011 May;13(5):634-9. doi:

Screening for CVD Ri sk Wi th ECG 98 RTI–UNC EPC Appendix C. Excluded Studies

10.1093/europace/eur016. PMID: 21325345. 273. Makimoto H, Nakagawa E, Takaki H, et al. Exclusion Code: X2. Augmented ST-segment elevation during 266. Macfarlane PW, Norrie J, Committee WE. recovery from exercise predicts cardiac The value of the electrocardiogram in risk events in patients with Brugada syndrome. J assessment in primary prevention: Am Coll Cardiol. 2010 Nov 2;56(19):1576- experience from the West of Scotland 84. doi: 10.1016/j.jacc.2010.06.033. PMID: Coronary Prevention Study. J 21029874. Exclusion Code: X2. Electrocardiol. 2007 Jan;40(1):101-9. doi: 274. Mandyam MC, Soliman EZ, Alonso A, et al. 10.1016/j.jelectrocard.2006.05.003. PMID: The QT interval and risk of incident atrial 17069838. Exclusion Code: X4. fibrillation. Heart Rhythm. 2013 267. Machado DB, Crow RS, Boland LL, et al. Oct;10(10):1562-8. doi: Electrocardiographic findings and incident 10.1016/j.hrthm.2013.07.023. PMID: coronary heart disease among participants in 23872693. Exclusion Code: X2. the Atherosclerosis Risk in Communities 275. Mandyam MC, Soliman EZ, Heckbert SR, (ARIC) study. Am J Cardiol. 2006 Apr et al. Long-term outcomes of left anterior 15;97(8):1176-81. doi: fascicular block in the absence of overt 10.1016/j.amjcard.2005.11.036. PMID: cardiovascular dis ease. Jama. 2013 Apr 16616022. Exclusion Code: X5. 17;309(15):1587-8. doi: 268. Magnani JW, Johnson VM, Sullivan LM, et 10.1001/jama.2013.2729. PMID: 23592102. al. P wave duration and risk of longitudinal Exclusion Code: X4. atrial fibrillation in persons >/= 60 years old 276. Marfella R, Sasso FC, Siniscalchi M, et al. (from the Framingham Heart Study). Am J Brief episodes of silent atrial fibrillation Cardiol. 2011 Mar 15;107(6):917-21.e1. predict clinical vascular brain disease in type doi: 10.1016/j.amjcard.2010.10.075. PMID: 2 diabetic patients. J Am Coll Cardiol. 2013 21255761. Exclusion Code: X5. Aug 6;62(6):525-30. doi: 269. Magnani JW, Wang N, Nelson KP, et al. 10.1016/j.jacc.2013.02.091. PMID: Electrocardiographic PR interval and 23684685. Exclusion Code: X4. adverse outcomes in older adults: the 277. Marine JE, Shetty V, Chow GV, et al. Health, Aging, and Body Composition Prevalence and prognostic significance of study. Circ Arrhythm Electrophysiol. 2013 exercise-induced nonsustained ventricular Feb;6(1):84-90. doi: tachycardia in asymptomatic volunteers: 10.1161/circep.112.975342. PMID: BLSA (Baltimore Longitudinal Study of 23243193. Exclusion Code: X2. Aging). J Am Coll Cardiol. 2013 Aug 270. Maheshwari A, Norby FL, Soliman EZ, et 13;62(7):595-600. doi: al. Relation of Prolonged P-Wave Duration 10.1016/j.jacc.2013.05.026. PMID: to Risk of Sudden Cardiac Death in the 23747767. Exclusion Code: X4. General Population (from the 278. Marrugat J, Subirana I, Comin E, et al. Atherosclerosis Risk in Communities Validity of an adaptation of the Framingham Study). Am J Cardiol. 2017 May cardiovascular risk function: the VERIFICA 01;119(9):1302-6. doi: Study. J Epidemiol Community Health. 2007 10.1016/j.amjcard.2017.01.012. PMID: Jan;61(1):40-7. doi: 28267962. Exclusion Code: X2. 10.1136/jech.2005.038505. PMID: 271. Mainous AG, 3rd, Everett CJ, Player MS, et 17183014. Exclusion Code: X4. al. Importance of a patient's personal health 279. Martin WH, 3rd, Xian H, Chandiramani P, history on assessments of future risk of et al. Cardiovascular mortality prediction in coronary heart disease. J Am Board Fam veterans with arm exercise vs Med. 2008 Sep-Oct;21(5):408-13. doi: pharmacologic myocardial perfusion 10.3122/jabfm.2008.05.080046. PMID: imaging. Am Heart J. 2015 Aug;170(2):362- 18772295. Exclusion Code: X3. 70. doi: 10.1016/j.ahj.2015.05.004. PMID: 272. Mainous AG, 3rd, Koopman RJ, Diaz VA, 26299235. Exclusion Code: X2. et al. A coronary heart disease risk score 280. Massing MW, Simpson RJ, Jr., Rautaharju based on patient-reported information. Am J PM, et al. Usefulness of ventricular Cardiol. 2007 May 1;99(9):1236-41. doi: premature complexes to predict coronary 10.1016/j.amjcard.2006.12.035. PMID: heart disease events and mortality (from the 17478150. Exclusion Code: X3. Atherosclerosis Risk In Communities cohort). Am J Cardiol. 2006 Dec

Screening for CVD Ri sk Wi th ECG 99 RTI–UNC EPC Appendix C. Excluded Studies

15;98(12):1609-12. doi: predicts adverse cardiac outcome at long- 10.1016/j.amjcard.2006.06.061. PMID: term follow-up. Coron Artery Dis. 2009 17145219. Exclusion Code: X5. Aug;20(5):337-42. doi: 281. May M, Lawlor DA, Brindle P, et al. 10.1097/MCA.0b013e32832c4589. PMID: Cardiovascular disease risk assessment in 19543085. Exclusion Code: X2. older women: can we improve on 289. Michaelides AP, Liakos CI, Vyssoulis GP, Framingham? British Women's Heart and et al. The interplay of exercise heart rate and Health prospective cohort study. Heart. blood pressure as a predictor of coronary 2006 Oct;92(10):1396-401. doi: artery disease and arterial hypertension. J 10.1136/hrt.2005.085381. PMID: 16547204. Clin Hypertens (Greenwich). 2013 Exclusion Code: X3. Mar;15(3):162-70. doi: 10.1111/jch.12035. 282. McAuley P, Pittsley J, Myers J, et al. Fitness PMID: 23458587. Exclusion Code: X4. and fatness as mortality predictors in healthy 290. Milne R, Gamble G, Whitlock G, et al. older men: the veterans exercise testing Discriminative ability of a risk-prediction study. J Gerontol A Biol Sci Med Sci. 2009 tool derived from the Framingham Heart Jun;64(6):695-9. doi: Study compared with single risk factors. N Z 10.1093/gerona/gln039. PMID: 19196639. Med J. 2003 Nov 07;116(1185):U663. Exclusion Code: X5. PMID: 14615805. Exclusion Code: X3. 283. McAuley PA, Sui X, Church TS, et al. The 291. Minkkinen M, Nieminen T, Verrier RL, et joint effects of cardiorespiratory fitness and al. Impaired exercise capacity predicts adiposity on mortality risk in men with sudden cardiac death in a low-risk hypertension. Am J Hypertens. 2009 population: enhanced specificity with Oct;22(10):1062-9. doi: heightened T-wave alternans. Ann Med. 10.1038/ajh.2009.122. PMID: 19617881. 2009;41(5):380-9. doi: Exclusion Code: X5. 10.1080/07853890902802971. PMID: 284. McEwan P, Williams JE, Griffiths JD, et al. 19301163. Exclusion Code: X5. Evaluating the performance of the 292. Miyamoto A, Hayashi H, Makiyama T, et al. Framingham risk equations in a population Risk determinants in individuals with a with diabetes. Diabet Med. 2004 spontaneous type 1 Brugada ECG. Circ J. Apr;21(4):318-23. doi: 10.1111/j.1464- 2011;75(4):844-51. PMID: 21343656. 5491.2004.01139.x. PMID: 15049932. Exclusion Code: X2. Exclusion Code: X3. 293. Miyasaka Y, Barnes ME, Gersh BJ, et al. 285. McNeill AM, Rosamond WD, Girman CJ, et Coronary ischemic events after first atrial al. The metabolic syndrome and 11-year risk fibrillation: risk and survival. Am J Med. of incident cardiovascular disease in the 2007 Apr;120(4):357-63. doi: atherosclerosis risk in communities study. 10.1016/j.amjmed.2006.06.042. PMID: Diabetes Care. 2005 Feb;28(2):385-90. 17398231. Exclusion Code: X2. PMID: 15677797. Exclusion Code: X3. 294. Moller CS, Haggstrom J, Zethelius B, et al. 286. Menotti A, Mulder I, Kromhout D, et al. The Age and follow-up time affect the association of silent electrocardiographic prognostic value of the ECG and findings with coronary deaths among elderly conventional cardiovascular risk factors for men in three European countries. The FINE stroke in adult men. J Epidemiol Community study. Acta Cardiol. 2001 Feb;56(1):27-36. Health. 2007 Aug;61(8):704-12. doi: doi: 10.2143/AC.56.1.2005590. PMID: 10.1136/jech.2006.048074. PMID: 11315121. Exclusion Code: X4. 17630370. Exclusion Code: X5. 287. Menotti A, Seccareccia F. 295. Mora S, Redberg RF, Cui Y, et al. Ability of Electrocardiographic Minnesota code exercise testing to predict cardiovascular findings predicting short-term mortality in and all-cause death in asymptomatic asymptomatic subjects. The Italian RIFLE women: a 20-year follow-up of the lipid Pooling Project (Risk Factors and Life research clinics prevalence study. JAMA. Expectancy). G Ital Cardiol. 1997 2003 Sep 24;290(12):1600-7. doi: Jan;27(1):40-9. PMID: 9199942. Exclusion 10.1001/jama.290.12.1600. PMID: Code: X5. 14506119. Exclusion Code: X5. 288. Michaelides AP, Andrikopoulos GK, 296. Mora S, Redberg RF, Sharrett AR, et al. Antoniades C, et al. Duration of treadmill Enhanced risk assessment in asymptomatic exercise testing combined with QRS score individuals with exercise testing and

Screening for CVD Ri sk Wi th ECG 100 RTI–UNC EPC Appendix C. Excluded Studies

Framingham risk scores. Circulation. 2005 doi: 10.1161/circulationaha.111.070045. Sep 13;112(11):1566-72. doi: PMID: 22308300. Exclusion Code: X4. 10.1161/CIRCULATIONAHA.105.542993. 304. Narayanan K, Zhang L, Kim C, et al. QRS PMID: 16144993. Exclusion Code: X5. fragmentation and sudden cardiac death in 297. Morin DP, Saad MN, Shams OF, et al. the obese and overweight. J Am Heart Relationships between the T-peak to T-end Assoc. 2015 Mar;4(3):e001654. doi: interval, ventricular tachyarrhythmia, and 10.1161/jaha.114.001654. PMID: 25762804. death in left ventricular systolic dysfunction. Exclusion Code: X7. Europace. 2012 Aug;14(8):1172-9. doi: 305. Nehme Z, Boyle MJ, Brown T. Diagnostic 10.1093/europace/eur426. PMID: 22277646. accuracy of prehospital clinical prediction Exclusion Code: X2. models to identify short-term outcomes in 298. Morshedi-Meibodi A, Larson MG, Levy D, patients with acute coronary syndromes: a et al. Heart rate recovery after treadmill systematic review (Provisional abstract). exercise testing and risk of cardiovascular Journal of Emergency Medicine. disease events (The Framingham Heart 2013;44(5):946-54.e6. PMID: DARE- Study). Am J Cardiol. 2002 Oct 12013030041. Exclusion Code: X2. 15;90(8):848-52. PMID: 12372572. 306. Nielsen JB, Kuhl JT, Pietersen A, et al. P- Exclusion Code: X4. wave duration and the risk of atrial 299. Moyer VA. Screening for coronary heart fibrillation: Results from the Copenhagen disease with electrocardiography: U.S. ECG Study. Heart Rhythm. 2015 Preventive Services Task Force Sep;12(9):1887-95. doi: recommendation statement. Ann Intern Med. 10.1016/j.hrthm.2015.04.026. PMID: 2012 Oct 2;157(7):512-8. doi: 25916567. Exclusion Code: X5. 10.7326/0003-4819-157-7-201210020- 307. Niemeijer MN, Berg ME, Eijgelsheim M, et 00514. PMID: 22847227. Exclusion Code: al. Short-term QT variability markers for the X7. prediction of ventricular arrhythmias and 300. Muramoto D, Yong CM, Singh N, et al. sudden cardiac death: a systematic review Patterns and prognosis of all components of (Provisional abstract). Database of Abstracts the J-wave pattern in multiethnic athletes of Reviews of Effects. 2014(2):1831-6. and ambulatory patients. Am Heart J. 2014 PMID: DARE-12014050251. Exclusion Feb;167(2):259-66. doi: Code: X5. 10.1016/j.ahj.2013.10.027. PMID: 308. Nieminen T, Verrier RL, Leino J, et al. 24439988. Exclusion Code: X7. Atrioventricular conduction and 301. Muzzarelli S, Pfisterer ME, Muller-Brand J, cardiovascular mortality: assessment of et al. Gate-keeper to coronary angiography: recovery PR interval is superior to pre- comparison of exercise testing, myocardial exercise measurement. Heart Rhythm. 2010 perfusion SPECT and individually tailored Jun;7(6):796-801. doi: approach for risk stratification. Int J 10.1016/j.hrthm.2010.02.029. PMID: Cardiovasc Imaging. 2010 Dec;26(8):871-9. 20188862. Exclusion Code: X2. doi: 10.1007/s10554-010-9627-y. PMID: 309. Noseworthy PA, Peloso GM, Hwang SJ, et 20411429. Exclusion Code: X2. al. QT interval and long-term mortality risk 302. Nakamura K, Berry NC, An PG, et al. in the Framingham Heart Study. Ann Significance of ST-segment elevation in Noninvasive Electrocardiol. 2012 lead aVR. Arch Intern Med. 2012 Mar Oct;17(4):340-8. doi: 10.1111/j.1542- 12;172(5):389-91. doi: 474X.2012.00535.x. PMID: 23094880. 10.1001/archinternmed.2011.2224. PMID: Exclusion Code: X4. 22412103. Exclusion Code: X7. 310. Ofoma U, He F, Shaffer ML, et al. 303. Nakamura Y, Okamura T, Higashiyama A, Premature cardiac contractions and risk of et al. Prognostic values of clockwise and incident ischemic stroke. J Am Heart Assoc. counterclockwise rotation for cardiovascular 2012 Oct;1(5):e002519. doi: mortality in Japanese subjects: a 24-year 10.1161/jaha.112.002519. PMID: 23316293. follow-up of the National Integrated Project Exc lusion Code: X5. for Prospective Observation of 311. Ofstad AP, Gullestad L, Orvik E, et al. Noncommunicable Disease and Its Trends in Interleukin-6 and activin A are the Aged, 1980-2004 (NIPPON DATA80). independently associated with Circulation. 2012 Mar 13;125(10):1226-33. cardiovascular events and mortality in type 2

Screening for CVD Ri sk Wi th ECG 101 RTI–UNC EPC Appendix C. Excluded Studies

diabetes: the prospective Asker and Baerum Intervention for End-Point Reduction in Cardiovascular Diabetes (ABCD) cohort Hypertension Study (LIFE). Circulation. study. Cardiovasc Diabetol. 2013;12:126. 2009 Apr 14;119(14):1883-91. doi: doi: 10.1186/1475-2840-12-126. PMID: 10.1161/circulationaha.108.812313. PMID: 23987834. Exclusion Code: X2. 19332468. Exclusion Code: X5. 312. Ogliari G, Mahinrad S, Stott DJ, et al. 320. Okin PM, Oikarinen L, Viitasalo M, et al. Resting heart rate, heart rate variability and Serial assessment of the functional decline in old age. Cmaj. 2015 electrocardiographic for Oct 20;187(15):E442-9. doi: prediction of new-onset heart failure during 10.1503/cmaj.150462. PMID: 26323697. antihypertensive treatment: the LIFE study. Exclusion Code: X2. Eur J Heart Fail. 2011 Apr;13(4):384-91. 313. Ohtomo K, Shigeeda T, Hirose A, et al. doi: 10.1093/eurjhf/hfq224. PMID: Silent myocardial ischaemia in patients with 21239405. Exclusion Code: X5. diabetic retinopathy. Acta Ophthalmol. 2014 321. Oliveira RB, Myers J, Araujo CG, et al. Sep;92(6):e492-3. doi: 10.1111/aos.12362. Maximal exercise oxygen as a PMID: 24698540. Exclusion Code: X4. predictor of mortality among male veterans 314. Okin PM, Anderson KM, Levy D, et al. referred for exercise testing. Eur J Heart rate adjustment of exercise-induced Cardiovasc Prev Rehabil. 2009 ST segment depression. Improved risk Jun;16(3):358-64. doi: stratification in the Framingham Offspring 10.1097/HJR.0b013e3283292fe8. PMID: Study. Circulation. 1991 Mar;83(3):866-74. 19357518. Exclusion Code: X3. PMID: 1999037. Exclusion Code: X5. 322. Olson KA, Viera AJ, Soliman EZ, et al. 315. Okin PM, Grandits G, Rautaharju PM, et al. Long-term prognosis associated with J-point Prognostic value of heart rate adjustment of elevation in a large middle-aged biracial exercise-induced ST segment depression in cohort: the ARIC study. Eur Heart J. 2011 the multiple risk factor intervention trial. J Dec;32(24):3098-106. doi: Am Coll Cardiol. 1996 May;27(6):1437-43. 10.1093/eurheartj/ehr264. PMID: 21785106. PMID: 8626955. Exclusion Code: X5. Exclusion Code: X5. 316. Okin PM, Hille DA, Kjeldsen SE, et al. 323. O'Neal WT, Shah AJ, Efird JT, et al. Persistence of left ventricular hypertrophy is Subclinical myocardial injury identified by associated with increased cardiovascular cardiac infarction/injury score and the risk morbidity and mortality in hypertensive of mortality in men and women free of patients with lower achieved systolic cardiovascular disease. Am J Cardiol. 2014 pressure during antihypertensive treatment. Oct 1;114(7):1018-23. doi: Blood Press. 2014 Apr;23(2):71-80. doi: 10.1016/j.amjcard.2014.06.032. PMID: 10.3109/08037051.2013.791414. PMID: 25129878. Exclusion Code: X5. 23721506. Exclusion Code: X2. 324. O'Neal WT, Wang YG, Wu HT, et al. 317. Okin PM, Kjeldsen SE, Julius S, et al. Effect Electrocardiographic J Wave and of changing heart rate during treatment of Cardiovascular Outcomes in the General hypertension on incidence of heart failure. Population (from the Atherosclerosis Risk In Am J Cardiol. 2012 Mar 1;109(5):699-704. Communities Study). Am J Cardiol. 2016 doi: 10.1016/j.amjcard.2011.10.026. PMID: Sep 15;118(6):811-5. doi: 22154318. Exclusion Code: X5. 10.1016/j.amjcard.2016.06.047. PMID: 318. Okin PM, Kjeldsen SE, Julius S, et al. All- 27596326. Exclusion Code: X5. cause and cardiovascular mortality in 325. Orford JL, Sesso HD, Stedman M, et al. A relation to changing heart rate during comparison of the Framingham and treatment of hypertensive patients with European Society of Cardiology coronary electrocardiographic left ventricular heart disease risk prediction models in the hypertrophy. Eur Heart J. 2010 normative aging study. Am Heart J. 2002 Sep;31(18):2271-9. doi: Jul;144(1):95-100. PMID: 12094194. 10.1093/eurheartj/ehq225. PMID: Exclusion Code: X3. 20601389. Exclusion Code: X5. 326. Ortega FB, Lee DC, Katzmarzyk PT, et al. 319. Okin PM, Oikarinen L, Viitasalo M, et al. The intriguing metabolically healthy but Prognostic value of changes in the obese phenotype: cardiovascular prognosis electrocardiographic strain pattern during and role of fitness. Eur Heart J. 2013 antihypertensive treatment: the Losartan Feb;34(5):389-97. doi:

Screening for CVD Ri sk Wi th ECG 102 RTI–UNC EPC Appendix C. Excluded Studies

10.1093/eurheartj/ehs174. PMID: 22947612. cardiometabolic risk factors and R-R Exclusion Code: X3. interval at 3years in a healthy population: 327. Ota C, Shiono SN, Fujino Y, et al. the RISC Study. Diabetes Metab. 2013 Prevalence and Prognostic Value of Early Sep;39(4):330-6. doi: Repolarization in Low Risk Surgical 10.1016/j.diabet.2013.05.008. PMID: Patients. Biomed Res Int. 23876398. Exclusion Code: X5. 2015;2015:309260. doi: 335. Payne CJ, Payne AR, Gibson SC, et al. Is 10.1155/2015/309260. PMID: 26266254. there still a role for preoperative 12-lead Exclusion Code: X4. electrocardiography? World J Surg. 2011 328. Ozgur ES, Nayci SA, Ozge C, et al. An Dec;35(12):2611-6. doi: 10.1007/s00268- integrated index combined by dynamic 011-1289-y. PMID: 21989644. Exclusion hyperinflation and exercise capacity in the Code: X2. prediction of morbidity and mortality in 336. Paynter NP, Chasman DI, Buring JE, et al. COPD. Respir Care. 2012 Sep;57(9):1452- Cardiovascular disease risk prediction with 9. doi: 10.4187/respcare.01440. PMID: and without knowledge of genetic variation 22348294. Exclusion Code: X4. at chromosome 9p21.3. Ann Intern Med. 329. Paisible AL, Chang CC, So-Armah KA, et 2009 Jan 20;150(2):65-72. PMID: al. HIV infection, cardiovascular disease 19153409. Exclusion Code: X3. risk factor profile, and risk for acute 337. Peel JB, Sui X, Matthews CE, et al. myocardial infarction. J Acquir Immune Cardiorespiratory fitness and digestive Defic Syndr. 2015 Feb 1;68(2):209-16. doi: cancer mortality: findings from the aerobics 10.1097/qai.0000000000000419. PMID: center longitudinal study. Cancer Epidemiol 25588033. Exclusion Code: X3. Biomarkers Prev. 2009 Apr;18(4):1111-7. 330. Palmerini T, Dangas G, Mehran R, et al. doi: 10.1158/1055-9965.epi-08-0846. Predictors and implications of stent PMID: 19293313. Exclusion Code: X5. thrombosis in non-ST-segment elevation 338. Perelshtein Brezinov O, Kivity S, Segev S, acute coronary syndromes: the ACUITY et al. Gender-Related Cardiovascular Risk in Trial. Circ Cardiovasc Interv. 2011 Dec Healthy Middle-Aged Adults. Am J Cardiol. 1;4(6):577-84. doi: 2016 Dec 01;118(11):1669-73. doi: 10.1161/circinterventions.111.963884. 10.1016/j.amjcard.2016.08.045. PMID: PMID: 22028471. Exclusion Code: X2. 27737731. Exclusion Code: X4. 331. Parikh NI, Gona P, Larson MG, et al. Long- 339. Perez-Rodon J, Martinez-Alday J, Baron- term trends in myocardial infarction Esquivias G, et al. Prognostic value of the incidence and case fatality in the National electrocardiogram in patients with syncope: Heart, Lung, and Blood Institute's data from the group for syncope study in the Framingham Heart study. Circulation. 2009 emergency room (GESINUR). Heart Mar 10;119(9):1203-10. doi: Rhythm. 2014 Nov;11(11):2035-44. doi: 10.1161/circulationaha.108.825364. PMID: 10.1016/j.hrthm.2014.06.037. PMID: 19237656. Exclusion Code: X5. 24993462. Exclusion Code: X2. 332. Park JI, Shin SY, Park SK, et al. Usefulness 340. Perino AC, Soofi M, Singh N, et al. The of the integrated scoring model of treadmill long-term prognostic value of the Q wave tests to predict myocardial ischemia and criteria for prior myocardial infarction silent myocardial ischemia in community- recommended in the universal definition of dwelling adults (from the Rancho Bernardo myocardial infarction. J Electrocardiol. study). Am J Cardiol. 2015 Apr 2015 Sep-Oct;48(5):798-802. doi: 15;115(8):1049-55. doi: 10.1016/j.jelectrocard.2015.07.004. PMID: 10.1016/j.amjcard.2015.01.536. PMID: 26233646. Exclusion Code: X4. 25728643. Exclusion Code: X5. 341. Peters RK, Cady LD, Jr., Bischoff DP, et al. 333. Park YM, Sui X, Liu J, et al. The effect of Physical fitness and subsequent myocardial cardiorespiratory fitness on age-related infarction in healthy workers. JAMA. 1983 lipids and lipoproteins. J Am Coll Cardiol. Jun 10;249(22):3052-6. PMID: 6854827. 2015 May 19;65(19):2091-100. doi: Exclusion Code: X5. 10.1016/j.jacc.2015.03.517. PMID: 342. Pfister R, Cairns R, Erdmann E, et al. 25975472. Exclusion Code: X5. Prognostic impact of electrocardiographic 334. Pataky Z, Golay A, Laville M, et al. Fasting signs in patients with Type 2 diabetes and at baseline influences the number of cardiovascular disease: results from the

Screening for CVD Ri sk Wi th ECG 103 RTI–UNC EPC Appendix C. Excluded Studies

PROactive study. Diabet Med. 2011 10.1016/j.amjcard.2014.04.005. PMID: Oct;28(10):1206-12. doi: 10.1111/j.1464- 24819898. Exclusion Code: X5. 5491.2011.03281.x. PMID: 21388447. 350. Qureshi WT, Alirhayim Z, Blaha MJ, et al. Exclusion Code: X2. Cardiorespiratory Fitness and Risk of 343. Pittaras AM, Faselis C, Doumas M, et al. Incident Atrial Fibrillation: Results From the Heart rate at rest, exercise capacity, and Henry Ford Exercise Testing (FIT) Project. mortality risk in veterans. Am J Cardiol. Circulation. 2015 May 26;131(21):1827-34. 2013 Nov 15;112(10):1605-9. doi: doi: 10.1161/circulationaha.114.014833. 10.1016/j.amjcard.2013.07.042. PMID: PMID: 25904645. Exclusion Code: X2. 24035162. Exclusion Code: X2. 351. Rafique AM, Biner S, Ray I, et al. Meta- 344. Pontone G, Andreini D, Bartorelli AL, et al. analysis of prognostic value of stress testing A long-term prognostic value of CT in patients with asymptomatic severe aortic angiography and exercise ECG in patients stenosis. Am J Cardiol. 2009 Oct with suspected CAD. JACC Cardiovasc 1;104(7):972-7. doi: Imaging. 2013 Jun;6(6):641-50. doi: 10.1016/j.amjcard.2009.05.044. PMID: 10.1016/j.jcmg.2013.01.015. PMID: 19766766. Exclusion Code: X2. 23764093. Exclusion Code: X2. 352. Rai M, Baker WL, Parker MW, et al. Meta- 345. Porthan K, Viitasalo M, Jula A, et al. analysis of optimal risk stratification in Predictive value of electrocardiographic QT patients >65 years of age. Am J Cardiol. interval and T-wave morphology parameters 2012 Oct 15;110(8):1092-9. doi: for all-cause and cardiovascular mortality in 10.1016/j.amjcard.2012.05.048. PMID: a general population sample. Heart Rhythm. 22795509. Exclusion Code: X7. 2009 Aug;6(8):1202-8, 8.e1. doi: 353. Rangel MO, O'Neal WT, Soliman EZ. 10.1016/j.hrthm.2009.05.006. PMID: Usefulness of the Electrocardiographic P- 19632634. Exclusion Code: X5. Wave Axis as a Predictor of Atrial 346. Prineas RJ, Grandits G, Rautaharju PM, et Fibrillation. Am J Cardiol. 2016 Jan al. Long-term prognostic significance of 1;117(1):100-4. doi: isolated minor electrocardiographic T-wave 10.1016/j.amjcard.2015.10.013. PMID: abnormalities in middle-aged men free of 26552511. Exclusion Code: X2. clinical cardiovascular disease (The Multiple 354. Rassi CH, Churchill TW, Tavares CA, et al. Risk Factor Intervention Trial [MRFIT]). Use of imaging and clinical data to screen Am J Cardiol. 2002 Dec 15;90(12):1391-5. for cardiovascular disease in asymptomatic PMID: 12480053. Exclusion Code: X5. diabetics. Cardiovasc Diabetol. 2016 Feb 347. Prineas RJ, Rautaharju PM, Grandits G, et 09;15:28. doi: 10.1186/s12933-016-0334-4. al. Independent risk for cardiovascular PMID: 26861208. Exclusion Code: X4. disease predicted by modified continuous 355. Rautaharju PM, Ge S, Nelson JC, et al. score electrocardiographic criteria for 6-year Comparison of mortality risk for incidence and regression of left ventricular electrocardiographic abnormalities in men hypertrophy among clinically disease free and women with and without coronary heart men: 16-year follow-up for the multiple risk disease (from the Cardiovascular Health factor intervention trial. J Electrocardiol. Study). Am J Cardiol. 2006 Feb 2001 Apr;34(2):91-101. PMID: 11320456. 1;97(3):309-15. doi: Exclusion Code: X5. 10.1016/j.amjcard.2005.08.046. PMID: 348. Puddu PE, Menotti A, Tolonen H, et al. 16442387. Exclusion Code: X5. Determinants of 40-year all-cause mortality 356. Rautaharju PM, Kooperberg C, Larson JC, in the European cohorts of the Seven et al. Electrocardiographic abnormalities Countries Study. Eur J Epidemiol. 2011 that predict coronary heart disease events Aug;26(8):595-608. doi: 10.1007/s10654- and mortality in postmenopausal women: 011-9600-7. PMID: 21713523. Exclusion the Women's Health Initiative. Circulation. Code: X5. 2006 Jan 31;113(4):473-80. doi: 349. Qureshi W, Shah AJ, Salahuddin T, et al. 10.1161/CIRCULATIONAHA.104.496091. Long-term mortality risk in individuals with PMID: 16449726. Exclusion Code: X5. atrial or ventricular premature complexes 357. Rautaharju PM, Kooperberg C, Larson JC, (results from the Third National Health and et al. Electrocardiographic predictors of Nutrition Examination Survey). Am J incident congestive heart failure and all- Cardiol. 2014 Jul 1;114(1):59-64. doi: cause mortality in postmenopausal women:

Screening for CVD Ri sk Wi th ECG 104 RTI–UNC EPC Appendix C. Excluded Studies

the Women's Health Initiative. Circulation. 10.1016/j.amjcard.2009.12.061. PMID: 2006 Jan 31;113(4):481-9. doi: 20451679. Exclusion Code: X3. 10.1161/CIRCULATIONAHA.105.537415. 365. Rich JD, Thenappan T, Freed B, et al. QTc PMID: 16449727. Exclusion Code: X5. prolongation is associated with impaired 358. Rautaharju PM, Prineas RJ, Eifler WJ, et al. right ventricular function and predicts Prognostic value of exercise mortality in . Int J electrocardiogram in men at high risk of Cardiol. 2013 Aug 10;167(3):669-76. doi: future coronary heart disease: Multiple Risk 10.1016/j.ijcard.2012.03.071. PMID: Factor Intervention Trial experience. J Am 22459397. Exclusion Code: X4. Coll Cardiol. 1986 Jul;8(1):1-10. PMID: 366. Ridker PM, Buring JE, Rifai N, et al. 3711503. Exclusion Code: X5. Development and validation of improved 359. Rautaharju PM, Zhang ZM, Haisty WK, Jr., algorithms for the assessment of global et al. Electrocardiographic predictors of cardiovascular risk in women: the Reynolds incident heart failure in men and women Risk Score. JAMA. 2007 Feb 14;297(6):611- free from manifest cardiovascular disease 9. doi: 10.1001/jama.297.6.611. PMID: (from the Atherosclerosis Risk in 17299196. Exclusion Code: X3. Communities [ARIC] study). Am J Cardiol. 367. Ridker PM, Paynter NP, Rifai N, et al. C- 2013 Sep 15;112(6):843-9. doi: reactive protein and parental history 10.1016/j.amjcard.2013.05.011. PMID: improve global cardiovascular risk 23768456. Exclusion Code: X4. prediction: the Reynolds Risk Score for 360. Rautaharju PM, Zhang ZM, Vitolins M, et men. Circulation. 2008 Nov al. Electrocardiographic repolarization- 25;118(22):2243-51, 4p following 51. doi: related variables as predictors of coronary 10.1161/CIRCULATIONAHA.108.814251. heart disease death in the women's health PMID: 18997194. Exclusion Code: X3. initiative study. J Am Heart Assoc. 2014 368. Rizk DV, Gutierrez O, Levitan EB, et al. Aug;3(4)doi: 10.1161/jaha.114.001005. Prevalence and prognosis of unrecognized PMID: 25074699. Exclusion Code: X5. myocardial in chronic kidney 361. Rautaharju PM, Zhang ZM, Warren J, et al. disease. Nephrol Dial Transplant. 2012 Electrocardiographic predictors of coronary Sep;27(9):3482-8. doi: 10.1093/ndt/gfr684. heart disease and sudden cardiac deaths in PMID: 22167594. Exclusion Code: X2. men and women free from cardiovascular 369. Rodrigues TC, Ehrlich J, Hunter CM, et al. disease in the Atherosclerosis Risk in Reduced heart rate variability predicts Communities study. J Am Heart Assoc. 2013 progression of coronary artery calcification Jun;2(3):e000061. doi: in adults with type 1 diabetes and controls 10.1161/jaha.113.000061. PMID: 23723252. without diabetes. Diabetes Technol Ther. Exclusion Code: X4. 2010 Dec;12(12):963-9. doi: 362. Ravera M, Cannavo R, Noberasco G, et al. 10.1089/dia.2010.0070. PMID: 21128843. High performance of a risk calculator that Exclusion Code: X5. includes renal function in predicting 370. Rollin A, Maury P, Kee F, et al. Isolated mortality of hypertensive patients in clinical negative T waves in the general population application. J Hypertens. 2014 is a powerful predicting factor of cardiac Jun;32(6):1245-54. doi: mortality and coronary heart disease. Int J 10.1097/hjh.0000000000000177. PMID: Cardiol. 2016 Jan 15;203:318-24. doi: 24751593. Exclusion Code: X3. 10.1016/j.ijcard.2015.10.118. PMID: 363. Reissigova J, Zvarova J. The Framingham 26523363. Exclusion Code: X5. risk function underestimated absolute 371. Rosengarten JA, Scott PA, Morgan JM. coronary heart disease risk in Czech men. Fragmented QRS for the prediction of Methods Inf Med. 2007;46(1):43-9. PMID: sudden cardiac death: a meta-analysis 17224979. Exclusion Code: X3. (Provisional abstract). Database of Abstracts 364. Rich JD, Chen S, Ward RP. Comparison of of Reviews of Effects. 2014(2):epub. PMID: high risk stress myocardial perfusion DARE-12014061825. Exclusion Code: X5. imaging findings in men with rapid versus 372. Rowin EJ, Maron BJ, Appelbaum E, et al. prolonged recovery of ST-s egment Significance of false negative depression after exercise stress testing. Am J electrocardiograms in preparticipation Cardiol. 2010 May 15;105(10):1361-4. doi: screening of athletes for hypertrophic . Am J Cardiol. 2012 Oct

Screening for CVD Ri sk Wi th ECG 105 RTI–UNC EPC Appendix C. Excluded Studies

1;110(7):1027-32. doi: Asians. J Am Heart Assoc. 2016 Mar 10.1016/j.amjcard.2012.05.035. PMID: 25;5(3):e002956. doi: 22809754. Exclusion Code: X2. 10.1161/jaha.115.002956. PMID: 27016575. 373. Rutter MK, Wahid ST, McComb JM, et al. Exclusion Code: X5. Significance of silent ischemia and 381. Savonen KP, Kiviniemi V, Laaksonen DE, microalbuminuria in predicting coronary et al. Two-minute heart rate recovery after events in asymptomatic patients with type 2 cycle ergometer exercise and all-cause diabetes. J Am Coll Cardiol. 2002 Jul mortality in middle-aged men. J Intern Med. 3;40(1):56-61. PMID: 12103256. Exclusion 2011 Dec;270(6):589-96. doi: Code: X5. 10.1111/j.1365-2796.2011.02434.x. PMID: 374. Rywik TM, O'Connor FC, Gittings NS, et al. 21801244. Exclusion Code: X4. Role of nondiagnostic exercise-induced ST- 382. Savonen KP, Lakka TA, Laukkanen JA, et segment abnormalities in predicting future al. Effectiveness of workload at the heart coronary events in asymptomatic volunteers. rate of 100 beats/min in predicting Circulation. 2002 Nov 26;106(22):2787-92. cardiovascular mortality in men aged 42, 48, PMID: 12451004. Exclusion Code: X5. 54, or 60 years at baseline. Am J Cardiol. 375. Rywik TM, Zink RC, Gittings NS, et al. 2007 Aug 15;100(4):563-8. doi: Independent prognostic significance of 10.1016/j.amjcard.2007.03.061. PMID: ischemic ST-segment response limited to 17697806. Exclusion Code: X5. recovery from treadmill exercise in 383. Schelbert EB, Cao JJ, Sigurdsson S, et al. asymptomatic subjects. Circulation. 1998 Prevalence and prognosis of unrecognized Jun 2;97(21):2117-22. PMID: 9626171. myocardial infarction determined by cardiac Exclusion Code: X5. magnetic resonance in older adults. Jama. 376. Saguner AM, Ganahl S, Baldinger SH, et al. 2012 Sep 5;308(9):890-6. doi: Usefulness of electrocardiographic 10.1001/2012.jama.11089. PMID: parameters for risk prediction in 22948699. Exclusion Code: X2. arrhythmogenic right ventricular dysplasia. 384. Scheltens T, Verschuren WM, Boshuizen Am J Cardiol. 2014 May 15;113(10):1728- HC, et al. Estimation of cardiovascular risk: 34. doi: 10.1016/j.amjcard.2014.02.031. a comparison between the Framingham and PMID: 24792740. Exclusion Code: X2. the SCORE model in people under 60 years 377. Salles GF, Cardoso CR, Fiszman R, et al. of age. Eur J Cardiovasc Prev Rehabil. Prognostic impact of baseline and serial 2008 Oct;15(5):562-6. doi: changes in electrocardiographic left 10.1097/HJR.0b013e3283063a65. PMID: ventricular hypertrophy in resistant 18756178. Exclusion Code: X3. hypertension. Am Heart J. 2010 385. Schinkel AF, Boiten HJ, van der Sijde JN, et May;159(5):833-40. doi: al. Prediction of 9-year cardiovascular 10.1016/j.ahj.2010.02.012. PMID: outcomes by myocardial perfusion imaging 20435193. Exclusion Code: X2. in patients with normal exercise 378. Salles GF, Cardoso CR, Fonseca LL, et al. electrocardiographic testing. Eur Heart J Prognostic significance of baseline heart rate Cardiovasc Imaging. 2012 Nov;13(11):900- and its interaction with beta-blocker use in 4. doi: 10.1093/ehjci/jes104. PMID: resistant hypertension: a cohort study. Am J 22588207. Exclusion Code: X2. Hypertens. 2013 Feb;26(2):218-26. doi: 386. Schinkel AF, Elhendy A, van Domburg RT, 10.1093/ajh/hps004. PMID: 23382406. et al. Prognostic significance of QRS Exclusion Code: X2. duration in patients with suspected coronary 379. Salles GF, Cardoso CR, Muxfeldt ES. artery disease referred for noninvasive Prognostic value of ventricular evaluation of myocardial ischemia. Am J repolarization prolongation in resistant Cardiol. 2009 Dec 1;104(11):1490-3. doi: hypertension: a prospective cohort study. J 10.1016/j.amjcard.2009.07.012. PMID: Hypertens. 2009 May;27(5):1094-101. doi: 19932780. Exclusion Code: X5. 10.1097/HJH.0b013e32832720b3. PMID: 387. Schlegel TT, Kulecz WB, Feiveson AH, et 19390353. Exclusion Code: X2. al. Accuracy of advanced versus strictly 380. Santhanakrishnan R, Wang N, Larson MG, conventional 12-lead ECG for detection and et al. Racial Differences in screening of coronary artery disease, left Electrocardiographic Characteristics and ventricular hypertrophy and left ventricular Prognostic Significance in Whites Versus systolic dysfunction. BMC Cardiovasc

Screening for CVD Ri sk Wi th ECG 106 RTI–UNC EPC Appendix C. Excluded Studies

Disord. 2010;10:28. doi: 10.1186/1471- 10.1111/1742-6723.12222. PMID: 2261-10-28. PMID: 20565702. Exclusion 24708007. Exclusion Code: X2. Code: X3. 395. Seicean S, Strohl KP, Seicean A, et al. Sleep 388. Schmidt C, Bergstrom G. Apolipoprotein disordered breathing as a risk of cardiac B/apolipoprotein A-I ratio and events in subjects with diabetes mellitus and apolipoprotein B: long-term predictors of normal exercise echocardiographic findings. myocardial infarction in initially healthy Am J Cardiol. 2013 Apr 15;111(8):1214-20. middle-aged men--a 13-year follow-up. doi: 10.1016/j.amjcard.2012.12.053. PMID: Angiology. 2014 Nov;65(10):901-5. doi: 23415514. Exclusion Code: X2. 10.1177/0003319713511849. PMID: 396. Sellers MB, Divers J, Lu L, et al. Prevalence 24277914. Exclusion Code: X3. and determinants of electrocardiographic 389. Schnabel RB, Rienstra M, Sullivan LM, et abnormalities in African Americans with al. Risk assessment for incident heart failure type 2 diabetes. J Epidemiol Glob Health. in individuals with atrial fibrillation. Eur J 2014 Dec;4(4):289-96. doi: Heart Fail. 2013 Aug;15(8):843-9. doi: 10.1016/j.jegh.2014.04.003. PMID: 10.1093/eurjhf/hft041. PMID: 23594831. 25455646. Exclusion Code: X5. Exclusion Code: X2. 397. Shah B, Bangalore S, Gianos E, et al. 390. Scholte AJ, Schuijf JD, Kharagjitsingh AV, Temporal trends in clinical characteristics of et al. Prevalence and predictors of an patients without known cardiovascular abnormal stress myocardial perfusion study disease with a first episode of myocardial in asymptomatic patients with type 2 infarction. Am Heart J. 2014 diabetes mellitus. Eur J Nucl Med Mol Apr;167(4):480-8.e1. doi: Imaging. 2009 Apr;36(4):567-75. doi: 10.1016/j.ahj.2013.12.019. PMID: 10.1007/s00259-008-0967-y. PMID: 24655696. Exclusion Code: X7. 18985347. Exclusion Code: X3. 398. Shah SA, Kambur T, Chan C, et al. Relation 391. Schouten O, van Kuijk JP, Flu WJ, et al. of short-term heart rate variability to Long-term outcome of prophylactic incident heart failure (from the Multi-Ethnic coronary revascularization in cardiac high- Study of Atherosclerosis). Am J Cardiol. risk patients undergoing major vascular 2013 Aug 15;112(4):533-40. doi: surgery (from the randomized DECREASE- 10.1016/j.amjcard.2013.04.018. PMID: V Pilot Study). Am J Cardiol. 2009 Apr 23683953. Exclusion Code: X4. 1;103(7):897-901. doi: 399. Shah SF, Meo SA. Usefulness of standard 10.1016/j.amjcard.2008.12.018. PMID: treadmill stress testing in women. J Pak Med 19327412. Exclusion Code: X2. Assoc. 2009 Apr;59(4):197-200. PMID: 392. Schroder K, Wegscheider K, Wenger NK, et 19402276. Exclusion Code: X2. al. Resting electrocardiogram predicts 400. Shaw LJ, Min JK, Budoff M, et al. Induced mortality in postmenopausal women with cardiovascular procedural costs and resource coronary heart disease or with risk factors consumption patterns after coronary artery for coronary heart disease. Eur J Prev calcium screening: results from the EISNER Cardiol. 2014 Jun;21(6):749-57. doi: (Early Identification of Subclinical 10.1177/2047487312454022. PMID: Atherosclerosis by Noninvasive Imaging 22752417. Exclusion Code: X2. Research) study. J Am Coll Cardiol. 2009 393. Schultz MG, Otahal P, Cleland VJ, et al. Sep 29;54(14):1258-67. doi: Exercise-induced hypertension, 10.1016/j.jacc.2009.07.018. PMID: cardiovascular events, and mortality in 19778667. Exclusion Code: X3. patients undergoing exercise stress testing: a 401. Shaw LJ, Wilson PW, Hachamovitch R, et systematic review and meta-analysis. Am J al. Improved near-term coronary artery Hypertens. 2013 Mar;26(3):357-66. doi: disease risk classification with gated stress 10.1093/ajh/hps053. PMID: 23382486. myocardial perfusion SPECT. JACC Exclusion Code: X7. Cardiovasc Imaging. 2010 Nov;3(11):1139- 394. Scott AC, Bilesky J, Lamanna A, et al. 48. doi: 10.1016/j.jcmg.2010.09.008. PMID: Limited utility of exercise stress testing in 21071002. Exclusion Code: X3. the evaluation of suspected acute coronary 402. Shaya GE, Al-Mallah MH, Hung RK, et al. syndrome in patients aged less than 40 years High Exercise Capacity Attenuates the Risk with intermediate risk features. Emerg Med of Early Mortality After a First Myocardial Australas. 2014 Apr;26(2):170-6. doi: Infarction: The Henry Ford Exercise Testing

Screening for CVD Ri sk Wi th ECG 107 RTI–UNC EPC Appendix C. Excluded Studies

(FIT) Project. Mayo Clin Proc. 2016 10.1177/2048872614548602. PMID: Feb;91(2):129-39. doi: 25228048. Exclusion Code: X2. 10.1016/j.mayocp.2015.11.012. PMID: 410. Simons LA, Simons J, Friedlander Y, et al. 26848000. Exclusion Code: X2. Risk functions for prediction of 403. Shibasaki K, Ogawa S, Yamada S, et al. cardiovascular disease in elderly Association of decreased sympathetic Australians: the Dubbo Study. Med J Aust. nervous activity with mortality of older 2003 Feb 3;178(3):113-6. PMID: 12558481. adults in long-term care. Geriatr Gerontol Exclusion Code: X3. Int. 2014 Jan;14(1):159-66. doi: 411. Sinner MF, Reinhard W, Muller M, et al. 10.1111/ggi.12074. PMID: 23879364. Association of early repolarization pattern Exclusion Code: X2. on ECG with risk of cardiac and all-cause 404. Shin SY, Park JI, Park SK, et al. Utility of mortality: a population-based prospective graded exercise tolerance tests for prediction cohort study (MONICA/KORA). PLoS Med. of cardiovascular mortality in old age: The 2010 Jul;7(7):e1000314. doi: Rancho Bernardo Study. Int J Cardiol. 2015 10.1371/journal.pmed.1000314. PMID: Feb 15;181:323-7. doi: 20668657. Exclusion Code: X5. 10.1016/j.ijcard.2014.12.026. PMID: 412. Siscovick DS, Ekelund LG, Johnson JL, et 25544200. Exclusion Code: X5. al. Sensitivity of exercise 405. Shiraishi J, Sawada T, Kimura S, et al. electrocardiography for acute cardiac events Enhanced cardiovascular protective effects during moderate and strenuous physical of valsartan in high-risk hypertensive activity. The Lipid Research Clinics patients with left ventricular hypertrophy-- Coronary Primary Prevention Trial. Arch sub-analysis of the KYOTO HEART study. Intern Med. 1991 Feb;151(2):325-30. Circ J. 2011;75(4):806-14. PMID: PMID: 1992960. Exclusion Code: X4. 21436597. Exclusion Code: X2. 413. Skalski J, Allison TG, Miller TD. The safety 406. Sigurdsson E, Sigfusson N, Sigvaldason H, of cardiopulmonary exercise testing in a et al. Silent ST-T changes in an population with high-risk cardiovascular epidemiologic cohort study--a marker of diseases. Circulation. 2012 Nov hypertension or coronary heart disease, or 20;126(21):2465-72. doi: both: the Reykjavik study. J Am Coll 10.1161/circulationaha.112.110460. PMID: Cardiol. 1996 Apr;27(5):1140-7. doi: 23091065. Exclusion Code: X2. 10.1016/0735-1097(95)00614-1. PMID: 414. Skretteberg PT, Grundvold I, Kjeldsen SE, 8609333. Exclusion Code: X5. et al. HDL-cholesterol and prediction of 407. Silventoinen K, Pankow J, Lindstrom J, et coronary heart disease: modified by physical al. The validity of the Finnish Diabetes Risk fitness? A 28-year follow-up of apparently Score for the prediction of the incidence of healthy men. Atherosclerosis. 2012 coronary heart disease and stroke, and total Jan;220(1):250-6. doi: mortality. Eur J Cardiovasc Prev Rehabil. 10.1016/j.atherosclerosis.2011.10.009. 2005 Oct;12(5):451-8. PMID: 16210931. PMID: 22062589. Exclusion Code: X3. Exclusion Code: X3. 415. Slattery ML, Jacobs DR, Jr. Physical fitness 408. Simmons RK, Sharp S, Boekholdt SM, et al. and cardiovascular disease mortality. The Evaluation of the Framingham risk score in US Railroad Study. Am J Epidemiol. 1988 the European Prospective Investigation of Mar;127(3):571-80. PMID: 3341361. Cancer-Norfolk cohort: does adding Exclusion Code: X5. improve the prediction 416. Smetana P, Schmidt A, Zabel M, et al. of coronary heart disease events? Arch Assessment of repolarization heterogeneity Intern Med. 2008 Jun 9;168(11):1209-16. for prediction of mortality in cardiovascular doi: 10.1001/archinte.168.11.1209. PMID: disease: peak to the end of the T wave 18541829. Exclusion Code: X3. interval and nondipolar repolarization 409. Simms AD, Weston CF, West RM, et al. components. J Electrocardiol. 2011 May- Mortality and missed opportunities along the Jun;44(3):301-8. doi: pathway of care for ST-elevation myocardial 10.1016/j.jelectrocard.2011.03.004. PMID: infarction: a national cohort study. Eur 21511064. Exclusion Code: X2. Heart J Acute Cardiovasc Care. 2015 417. Smith JG, Platonov PG, Hedblad B, et al. Jun;4(3):241-53. doi: Atrial fibrillation in the Malmo Diet and Cancer study: a study of occurrence, risk

Screening for CVD Ri sk Wi th ECG 108 RTI–UNC EPC Appendix C. Excluded Studies

factors and diagnostic validity. Eur J 425. Soliman EZ, Shah AJ, Boerkircher A, et al. Epidemiol. 2010 Feb;25(2):95-102. doi: Inter-relationship between 10.1007/s10654-009-9404-1. PMID: electrocardiographic left ventricular 19936945. Exclusion Code: X5. hypertrophy and QT prolongation as 418. Snoek JA, Jongman JK, Brandon T, et al. predictors of increased risk of mortality in Performance of the Lausanne questionnaire the general population. Circ Arrhythm and the 2010 European Society of Electrophysiol. 2014 Jun;7(3):400-6. doi: Cardiology criteria for ECG interpretation in 10.1161/circep.113.001396. PMID: athletes. Eur J Prev Cardiol. 2015 24762807. Exclusion Code: X5. Mar;22(3):397-405. doi: 426. Soliman EZ, Zhang ZM, Chen LY, et al. 10.1177/2047487313506827. PMID: Usefulness of Maintaining a Normal 24057687. Exclusion Code: X5. Electrocardiogram Over Time for Predicting 419. Soliman EZ, Cammarata M, Li Y. Cardiovascular Health. Am J Cardiol. 2017 Explaining the inconsistent associations of Jan 15;119(2):249-55. doi: PR interval with mortality: the role of P- 10.1016/j.amjcard.2016.09.051. PMID: duration contribution to the length of PR 28126148. Exclusion Code: X2. interval. Heart Rhythm. 2014 Jan;11(1):93- 427. Soofi M, Jain NA, Myers J, et al. A New 12- 8. doi: 10.1016/j.hrthm.2013.10.003. PMID: Lead ECG Prognostic Score. Ann 24096163. Exclusion Code: X2. Noninvasive Electrocardiol. 2015 420. Soliman EZ, Elsalam MA, Li Y. The Nov;20(6):554-60. doi: 10.1111/anec.12261. relationship between high resting heart rate PMID: 25640186. Exclusion Code: X4. and ventricular arrhythmogenesis in patients 428. Stacey RB, Leaverton PE, Schocken DD, et referred to ambulatory 24 h al. and the association with electrocardiographic recording. Europace. unrecognized myocardial infarction in the 2010 Feb;12(2):261-5. doi: multi-ethnic study of atherosclerosis. Am 10.1093/europace/eup344. PMID: Heart J. 2015 Nov;170(5):923-8. doi: 19887457. Exclusion Code: X2. 10.1016/j.ahj.2015.08.003. PMID: 421. Soliman EZ, Howard G, Cushman M, et al. 26542500. Exclusion Code: X3. Prolongation of QTc and risk of stroke: The 429. Stavrakis S, Patel N, Te C, et al. REGARDS (REasons for Geographic and Development and validation of a prognostic Racial Differences in Stroke) study. J Am index for risk stratification of patients with Coll Cardiol. 2012 Apr 17;59(16):1460-7. early repolarization. Ann Noninvasive doi: 10.1016/j.jacc.2012.01.025. PMID: Electrocardiol. 2012 Oct;17(4):361-71. doi: 22497826. Exclusion Code: X2. 10.1111/j.1542-474X.2012.00533.x. PMID: 422. Soliman EZ, Howard G, Meschia JF, et al. 23094882. Exclusion Code: X7. Self-reported atrial fibrillation and risk of 430. Stefanini GG, Kalesan B, Pilgrim T, et al. stroke in the Reasons for Geographic and Impact of sex on clinical and angiographic Racial Differences in Stroke (REGARDS) outcomes among patients undergoing study. Stroke. 2011 Oct;42(10):2950-3. doi: revascularization with drug-eluting stents. 10.1161/strokeaha.111.621367. PMID: JACC Cardiovasc Interv. 2012 21817138. Exclusion Code: X5. Mar;5(3):301-10. doi: 423. Soliman EZ, Lopez F, O'Neal WT, et al. 10.1016/j.jcin.2011.11.011. PMID: Atrial Fibrillation and Risk of ST-Segment- 22440496. Exclusion Code: X2. Elevation Versus Non-ST-Segment- 431. Stein PK, Barzilay JI, Chaves PH, et al. Elevation Myocardial Infarction: The Heart rate variability and its changes over 5 Atherosclerosis Risk in Communities years in older adults. Age Ageing. 2009 (ARIC) Study. Circulation. 2015 May Mar;38(2):212-8. doi: 26;131(21):1843-50. doi: 10.1093/ageing/afn292. PMID: 19147739. 10.1161/circulationaha.114.014145. PMID: Exclusion Code: X3. 25918127. Exclusion Code: X4. 432. Stein R, Nguyen P, Abella J, et al. 424. Soliman EZ, Prineas RJ. Early Prevalence and prognostic significance of repolarization and sudden cardiac death. exercise-induced right bundle branch block. South Med J. 2009 Jan;102(1):110-1. doi: Am J Cardiol. 2010 Mar 1;105(5):677-80. 10.1097/SMJ.0b013e31818fbe44. PMID: doi: 10.1016/j.amjcard.2009.10.050. PMID: 19077760. Exclusion Code: X7. 20185016. Exclusion Code: X2.

Screening for CVD Ri sk Wi th ECG 109 RTI–UNC EPC Appendix C. Excluded Studies

433. Stein R, Sallam K, Adhikarla C, et al. doi: 10.1016/j.atherosclerosis.2015.11.026. Natural history of early repolarization in the PMID: 26694693. Exclusion Code: X5. inferior leads. Ann Noninvasive 441. Sui X, LaMonte MJ, Blair SN. Electrocardiol. 2012 Oct;17(4):331-9. doi: Cardiorespiratory fitness as a predictor of 10.1111/j.1542-474X.2012.00537.x. PMID: nonfatal cardiovascular events in 23094879. Exclusion Code: X7. asymptomatic women and men. Am J 434. Stephens JW, Ambler G, Vallance P, et al. Epidemiol. 2007 Jun 15;165(12):1413-23. Cardiovascular risk and diabetes. Are the doi: 10.1093/aje/kwm031. PMID: methods of risk prediction satisfactory? Eur 17406007. Exclusion Code: X5. J Cardiovasc Prev Rehabil. 2004 442. Suka M, Sugimori H, Yoshida K. Dec;11(6):521-8. PMID: 15580065. Application of the updated Framingham risk Exclusion Code: X3. score to Japanese men. Hypertens Res. 2001 435. Stern MP, Williams K, Gonzalez- Nov;24(6):685-9. PMID: 11768728. Villalpando C, et al. Does the metabolic Exclusion Code: X3. syndrome improve identification of 443. Suka M, Sugimori H, Yoshida K. Validity of individuals at risk of type 2 diabetes and/or the Framingham risk model applied to cardiovascular disease? Diabetes Care. 2004 Japanese men. Methods Inf Med. Nov;27(11):2676-81. PMID: 15505004. 2002;41(3):213-5. PMID: 12162145. Exclusion Code: X3. Exclusion Code: X3. 436. Stewart RA, Young AA, Anderson C, et al. 444. Sun Z, Ng KH. Diagnostic value of coronary Relationship between QRS duration and left CT angiography with prospective ECG- ventricular mass and volume in patients at gating in the diagnosis of coronary artery high cardiovascular risk. Heart. 2011 disease: a systematic review and meta- Nov;97(21):1766-70. doi: 10.1136/heartjnl- analysis (Provisional abstract). International 2011-300297. PMID: 21835757. Exclusion Journal of Cardiovascular Imaging. Code: X5. 2012;28(8):2109-19. PMID: DARE- 437. Stiles MC, Seaquist ER, Yale JF, et al. Is 12013001989. Exclusion Code: X3. silent myocardial infarction more common 445. Sung J, Cho SJ, Choe YH, et al. in women with type 2 diabetes than in men? Relationship between aerobic fitness and J Diabetes Complications. 2012 Mar- progression of coronary atherosclerosis. Apr;26(2):118-22. doi: Heart Vessels. 2016 Sep;31(9):1418-23. doi: 10.1016/j.jdiacomp.2012.02.002. PMID: 10.1007/s00380-015-0745-2. PMID: 22446034. Exclusion Code: X5. 26400860. Exclusion Code: X5. 438. Strom Moller C, Zethelius B, Sundstrom J, 446. Sutherland SE, Gazes PC, Keil JE, et al. et al. Persistent ischaemic ECG Electrocardiographic abnormalities and 30- abnormalities on repeated ECG examination year mortality among white and black men have important prognostic value for of the Charleston Heart Study. Circulation. cardiovascular disease beyond established 1993 Dec;88(6):2685-92. PMID: 8252679. risk factors: a population-based study in Exclusion Code: X5. middle-aged men with up to 32 years of 447. Suvisaari J, Perala J, Saarni SI, et al. follow-up. Heart. 2007 Sep;93(9):1104-10. Coronary heart disease and cardiac doi: 10.1136/hrt.2006.109116. PMID: conduction abnormalities in persons with 17483125. Exclusion Code: X10. psychotic disorders in a general population. 439. Suadicani P, Hein HO, Gyntelberg F. Psychiatry Res. 2010 Jan 30;175(1-2):126- Antihypertensive treatment, high 32. doi: 10.1016/j.psychres.2008.07.021. triglycerides, and low high-density PMID: 19926142. Exclusion Code: X2. lipoprotein cholesterol and risk of ischemic 448. Svart K, Lehtinen R, Nieminen T, et al. heart disease mortality: a 16-year follow-up Exercise electrocardiography detection of in the Copenhagen male study. Metab Syndr coronary artery disease by ST-segment Relat Disord. 2010 Jun;8(3):215-22. doi: depression/heart rate hysteresis in women: 10.1089/met.2009.0072. PMID: 20156073. the Finnish Cardiovascular Study. Int J Exclusion Code: X5. Cardiol. 2010 Apr 15;140(2):182-8. doi: 440. Suh B, Park S, Shin DW, et al. Early 10.1016/j.ijcard.2008.11.038. PMID: repolarization is associated with significant 19068271. Exclusion Code: X4. coronary artery stenosis in asymptomatic 449. Tait J, Ashton T. Critical left main coronary adults. Atherosclerosis. 2016 Feb;245:50-3. artery stenosis identified by colour Doppler

Screening for CVD Ri sk Wi th ECG 110 RTI–UNC EPC Appendix C. Excluded Studies

screening during pre-exercise stress cardiovascular outcomes: Tehran Lipid and echocardiogram. Can J Cardiol. 2011 May- Glucose Study. Int J Cardiol. 2009 Mar Jun;27(3):390.e13-4. doi: 6;132(3):369-74. doi: 10.1016/j.cjca.2010.12.055. PMID: 10.1016/j.ijcard.2007.11.085. PMID: 21601777. Exclusion Code: X7. 18242731. Exclusion Code: X3. 450. Takase B, Masaki N, Hattori H, et al. 458. Tota-Maharaj R, Blaha MJ, Blankstein R, et Usefulness of automatic QT dispersion al. Association of coronary artery calcium measurement for detecting exercise-induced and coronary heart disease events in young myocardial ischemia. Anadolu Kardiyol and elderly participants in the multi-ethnic Derg. 2009 Jun;9(3):189-95. PMID: study of atherosclerosis: a secondary 19520652. Exclusion Code: X2. analysis of a prospective, population-based 451. Teodorescu C, Reinier K, Uy-Evanado A, et cohort. Mayo Clin Proc. 2014 al. Resting heart rate and risk of sudden Oct;89(10):1350-9. doi: cardiac death in the general population: 10.1016/j.mayocp.2014.05.017. PMID: influence of left ventricular systolic 25236430. Exclusion Code: X3. dysfunction and heart rate-modulating 459. Turakhia MP, Ullal AJ, Hoang DD, et al. drugs. Heart Rhythm. 2013 Aug;10(8):1153- Feasibility of extended ambulatory 8. doi: 10.1016/j.hrthm.2013.05.009. PMID: electrocardiogram monitoring to identify 23680897. Exclusion Code: X2. silent atrial fibrillation in high-risk patients: 452. Teodorescu C, Reinier K, Uy-Evanado A, et the Screening Study for Undiagnosed Atrial al. Prolonged QRS duration on the resting Fibrillation (STUDY-AF). Clin Cardiol. ECG is associated with sudden death risk in 2015 May;38(5):285-92. doi: coronary disease, independent of prolonged 10.1002/clc.22387. PMID: 25873476. ventricular repolarization. Heart Rhythm. Exclusion Code: X4. 2011 Oct;8(10):1562-7. doi: 460. Uberoi A, Sallam K, Perez M, et al. 10.1016/j.hrthm.2011.06.011. PMID: Prognostic implications of Q waves and T- 21699869. Exclusion Code: X2. wave inversion associated with early 453. Tereshchenko LG, Shah AJ, Li Y, et al. repolarization. Mayo Clin Proc. 2012 Electrocardiographic deep terminal Jul;87(7):614-9. doi: negativity of the P wave in V1 and risk of 10.1016/j.mayocp.2012.04.009. PMID: mortality: the National Health and Nutrition 22766081. Exclusion Code: X7. Examination Survey III. J Cardiovasc 461. Ulmer H, Kollerits B, Kelleher C, et al. Electrophysiol. 2014 Nov;25(11):1242-8. Predictive accuracy of the SCORE risk doi: 10.1111/jce.12453. PMID: 24837486. function for cardiovascular disease in Exclusion Code: X4. clinical practice: a prospective evaluation of 454. Tikkanen JT, Anttonen O, Junttila MJ, et al. 44 649 Austrian men and women. Eur J Long-term outcome associated with early Cardiovasc Prev Rehabil. 2005 repolarization on electrocardiography. N Oct;12(5):433-41. PMID: 16210929. Engl J Med. 2009 Dec 24;361(26):2529-37. Exclusion Code: X3. doi: 10.1056/NEJMoa0907589. PMID: 462. Uthamalingam S, Zheng H, Leavitt M, et al. 19917913. Exclusion Code: X5. Exercise-induced ST-segment elevation in 455. Tikkanen JT, Huikuri HV. Early ECG lead aVR is a useful indicator of repolarization ECG pattern in the Finnish significant left main or ostial LAD coronary general population. J Electrocardiol. 2013 artery stenosis. JACC Cardiovasc Imaging. Sep-Oct;46(5):439-41. doi: 2011 Feb;4(2):176-86. doi: 10.1016/j.jelectrocard.2013.06.012. PMID: 10.1016/j.jcmg.2010.11.014. PMID: 23871658. Exclusion Code: X7. 21329903. Exclusion Code: X2. 456. Tikkanen JT, Junttila MJ, Anttonen O, et al. 463. Vaara S, Nieminen MS, Lokki ML, et al. Early repolarization: electrocardiographic Cohort Profile: the Corogene study. Int J phenotypes associated with favorable long- Epidemiol. 2012 Oct;41(5):1265-71. doi: term outcome. Circulation. 2011 Jun 10.1093/ije/dyr090. PMID: 21642350. 14;123(23):2666-73. doi: Exclusion Code: X2. 10.1161/circulationaha.110.014068. PMID: 464. Vaidya D, Yanek LR, Moy TF, et al. 21632493. Exclusion Code: X5. Incidence of coronary artery disease in 457. Tohidi M, Hadaegh F, Harati H, et al. C- siblings of patients with premature coronary reactive protein in risk prediction of artery disease: 10 years of follow-up. Am J

Screening for CVD Ri sk Wi th ECG 111 RTI–UNC EPC Appendix C. Excluded Studies

Cardiol. 2007 Nov 1;100(9):1410-5. doi: 10.1111/anec.12231. PMID: 25393741. 10.1016/j.amjcard.2007.06.031. PMID: Exclusion Code: X5. 17950799. Exclusion Code: X3. 472. Vliegenthart R, Oudkerk M, Hofman A, et 465. Vakil KP, Malhotra S, Sawada S, et al. al. Coronary calcification improves Waist circumference and metabolic cardiovascular risk prediction in the elderly. syndrome: the risk for silent coronary artery Circulation. 2005 Jul 26;112(4):572-7. doi: disease in males. Metab Syndr Relat Disord. 10.1161/CIRCULATIONAHA.104.488916. 2012 Jun;10(3):225-31. doi: PMID: 16009800. Exclusion Code: X3. 10.1089/met.2011.0099. PMID: 22324791. 473. von Ziegler F, Greif M, Tittus J, et al. Exclusion Code: X7. Distribution of coronary calcifications in 466. van Noord C, Sturkenboom MC, Straus SM, patients with suspected coronary heart et al. Serum glucose and insulin are disease. Am Heart J. 2014 Apr;167(4):568- associated with QTc and RR intervals in 75. doi: 10.1016/j.ahj.2013.12.011. PMID: nondiabetic elderly. Eur J Endocrinol. 2010 24655707. Exclusion Code: X2. Feb;162(2):241-8. doi: 10.1530/eje-09-0878. 474. Waks JW, Sitlani CM, Soliman EZ, et al. PMID: 19897609. Exclusion Code: X5. Global Electric Heterogeneity Risk Score 467. Vandvik PO, Lincoff AM, Gore JM, et al. for Prediction of Sudden Cardiac Death in Primary and secondary prevention of the General Population: The Atherosclerosis cardiovascular disease: Antithrombotic Risk in Communities (ARIC) and Therapy and Prevention of Thrombosis, 9th Cardiovascular Health (CHS) Studies. ed: American College of Chest Physicians Circulation. 2016 Jun 07;133(23):2222-34. Evidence-Based Clinical Practice doi: 10.1161/circulationaha.116.021306. Guidelines. Chest. 2012 Feb;141(2 PMID: 27081116. Exclusion Code: X4. Suppl):e637S-68S. doi: 10.1378/chest.11- 475. Waks JW, Soliman EZ, Henrikson CA, et al. 2306. PMID: 22315274. Exclusion Code: Beat-to-beat spatiotemporal variability in the X7. T vector is associated with sudden cardiac 468. Vepsalainen T, Laakso M, Lehto S, et al. death in participants without left ventricular Prolonged P wave duration predicts stroke hypertrophy: the Atherosclerosis Risk in mortality among type 2 diabetic patients Communities (ARIC) Study. J Am Heart with prevalent non-major macrovascular Assoc. 2015 Jan;4(1):e001357. doi: disease. BMC Cardiovasc Disord. 10.1161/jaha.114.001357. PMID: 25600143. 2014;14:168. doi: 10.1186/1471-2261-14- Exclusion Code: X3. 168. PMID: 25425321. Exclusion Code: X5. 476. Walsh JA, 3rd, Ilkhanoff L, Soliman EZ, et 469. Verdecchia P, Reboldi G, Di Pasquale G, et al. Natural history of the early repolarization al. Prognostic usefulness of left ventricular pattern in a biracial cohort: CARDIA hypertrophy by electrocardiography in (Coronary Artery Risk Development in patients with atrial fibrillation (from the Young Adults) Study. J Am Coll Cardiol. Randomized Evaluation of Long-Term 2013 Feb 26;61(8):863-9. doi: Anticoagulant Therapy Study). Am J 10.1016/j.jacc.2012.11.053. PMID: Cardiol. 2014 Feb 15;113(4):669-75. doi: 23428218. Exclusion Code: X4. 10.1016/j.amjcard.2013.10.045. PMID: 477. Walsh JA, 3rd, Soliman EZ, Ilkhanoff L, et 24359765. Exclusion Code: X2. al. Prognostic value of frontal QRS-T angle 470. Vergnaud AC, Bertrais S, Galan P, et al. in patients without clinical evidence of Ten-year risk prediction in French men cardiovascular disease (from the Multi- using the Framingham coronary score: Ethnic Study of Atherosclerosis). Am J results from the national SU.VI.MAX Cardiol. 2013 Dec 15;112(12):1880-4. doi: cohort. Prev Med. 2008 Jul;47(1):61-5. doi: 10.1016/j.amjcard.2013.08.017. PMID: 10.1016/j.ypmed.2008.02.023. PMID: 24063831. Exclusion Code: X4. 18456313. Exclusion Code: X3. 478. Wang Z, Hoy WE. Is the Framingham 471. Vinsonneau U, Pangnarind-Hein z V, coronary heart disease absolute risk function Paleiron N, et al. Evolution of Early applicable to Aboriginal people? Med J Repolarization Patterns after 5 Years in a Aust. 2005 Jan 17;182(2):66-9. PMID: Military Population at Low Cardiovascular 15651963. Exclusion Code: X3. Risk and Practical Implications in Military 479. Wannamethee SG, Shaper AG, Lennon L, et Medical Expertise. Ann Noninvasive al. Metabolic syndrome vs Framingham Electrocardiol. 2015 Sep;20(5):420-5. doi: Risk Score for prediction of coronary heart

Screening for CVD Ri sk Wi th ECG 112 RTI–UNC EPC Appendix C. Excluded Studies

disease, stroke, and type 2 diabetes mellitus. interval to predict left ventricular diastolic Arch Intern Med. 2005 Dec 12- dysfunction. Am J Cardiol. 2011 Dec 26;165(22):2644-50. doi: 15;108(12):1760-6. doi: 10.1001/archinte.165.22.2644. PMID: 10.1016/j.amjcard.2011.07.050. PMID: 16344423. Exclusion Code: X3. 21907948. Exclusion Code: X2. 480. Watanabe H, Makita N, Tanabe N, et al. 488. Wilson PW, Bozeman SR, Burton TM, et al. Electrocardiographic abnormalities and risk Prediction of first events of coronary heart of complete atrioventricular block. Int J disease and stroke with consideration of Cardiol. 2012 Mar 22;155(3):462-4. doi: adiposity. Circulation. 2008 Jul 10.1016/j.ijcard.2011.12.028. PMID: 8;118(2):124-30. doi: 22225759. Exclusion Code: X5. 10.1161/CIRCULATIONAHA.108.772962. 481. Waure C, Cadeddu C, Gualano MR, et al. PMID: 18591432. Exclusion Code: X3. Telemedicine for the reduction of 489. Wilson PW, D'Agostino RB, Levy D, et al. myocardial infarction mortality: a systematic Prediction of coronary heart disease using review and a meta-analysis of published risk factor categories. Circulation. 1998 studies (Structured abstract). Telemedicine May 12;97(18):1837-47. PMID: 9603539. and e-Health. 2012;18(5):323-8. PMID: Exclusion Code: X3. DARE-12012049125. Exclusion Code: X2. 490. Wirawan IM, Aldington S, Griffiths RF, et 482. Weiner DE, Tighiouart H, Elsayed EF, et al. al. Cardiovascular investigations of airline The Framingham predictive instrument in pilots with excessive cardiovascular risk. chronic kidney disease. J Am Coll Cardiol. Aviat Space Environ Med. 2013 2007 Jul 17;50(3):217-24. doi: Jun;84(6):608-12. PMID: 23745289. 10.1016/j.jacc.2007.03.037. PMID: Exclusion Code: X4. 17631213. Exclusion Code: X3. 491. Woodward M, Brindle P, Tunstall-Pedoe H, 483. Welsh P, Doolin O, Willeit P, et al. N- et al. Adding social deprivation and family terminal pro-B-type natriuretic peptide and history to cardiovascular risk assessment: the prediction of primary cardiovascular the ASSIGN score from the Scottish Heart events: results from 15-year follow-up of Health Extended Cohort (SHHEC). Heart. WOSCOPS. Eur Heart J. 2013 2007 Feb;93(2):172-6. doi: Feb;34(6):443-50. doi: 10.1136/hrt.2006.108167. PMID: 17090561. 10.1093/eurheartj/ehs239. PMID: 22942340. Exc lu sion Code: X3. Exclusion Code: X3. 492. Worthington JM, Gattellari M, Leung DY. 484. Wen L, Sun ML, An P, et al. Frequency of 'Where there's ...': are premature supraventricular arrhythmias in patients with ventricular complexes a new risk factor for idiopathic pulmonary arterial hypertension. stroke? Stroke. 2010 Apr;41(4):572-3. doi: Am J Cardiol. 2014 Nov 1;114(9):1420-5. 10.1161/strokeaha.109.574426. PMID: doi: 10.1016/j.amjcard.2014.07.079. PMID: 20167909. Exclusion Code: X7. 25217453. Exclusion Code: X2. 493. Wu Y, Liu X, Li X, et al. Estimation of 10- 485. Wenger NK, Mischke JM, Schroeder R, et year risk of fatal and nonfatal ischemic al. Electrocardiograms of menopausal cardiovascular diseases in Chinese adults. women with coronary heart disease or at Circulation. 2006 Nov 21;114(21):2217-25. increased risk for its occurrence. Am J doi: Cardiol. 2010 Dec 1;106(11):1580-7. doi: 10.1161/CIRCULATIONAHA.105.607499. 10.1016/j.amjcard.2010.07.032. PMID: PMID: 17088464. Exclusion Code: X3. 21094358. Exclusion Code: X7. 494. Wu YT, Chien CL, Wang SY, et al. Gender 486. Whang W, Peacock J, Soliman EZ, et al. differences in myocardial perfusion defect in Relations between depressive symptoms, asymptomatic postmenopausal women and anxiety, and T Wave abnormalities in men with and without diabetes mellitus. J subjects without clinically-apparent Womens Health (Larchmt). 2013 cardiovascular disease (from the Multi- May;22(5):439-44. doi: Ethnic Study of Atherosclerosis [MESA]). 10.1089/jwh.2012.4055. PMID: 23600438. Am J Cardiol. 2014 Dec 15;114(12):1917- Exclusion Code: X4. 22. doi: 10.1016/j.amjcard.2014.09.034. 495. Xanthakis V, Enserro DM, Murabito JM, et PMID: 25438922. Exclusion Code: X5. al. Ideal cardiovascular health: associations 487. Wilcox JE, Rosenberg J, Vallakati A, et al. with biomarkers and subclinical disease and Usefulness of electrocardiographic QT impact on incidence of cardiovascular

Screening for CVD Ri sk Wi th ECG 113 RTI–UNC EPC Appendix C. Excluded Studies

disease in the Framingham Offspring Study. 502. Zhang Y, Protogerou AD, Iaria P, et al. Circulation. 2014 Nov 4;130(19):1676-83. Prognosis in the hospitalized very elderly: doi: 10.1161/circulationaha.114.009273. the PROTEGER study. Int J Cardiol. 2013 PMID: 25274000. Exclusion Code: X3. Oct 3;168(3):2714-9. doi: 496. Yanagisawa S, Miki K, Yasuda N, et al. The 10.1016/j.ijcard.2013.03.021. PMID: prognostic value of treadmill exercise 23578896. Exclusion Code: X2. testing in very elderly patients: heart rate 503. Zhang ZM, Prineas RJ, Case D, et al. recovery as a predictor of mortality in Gender differences between the Minnesota octogenarians. Europace. 2011 code and Novacode electrocardiographic Jan;13(1):114-20. doi: prognostication of coronary heart disease in 10.1093/europace/euq422. PMID: the cardiovascular health study. Am J 21084358. Exclusion Code: X2. Cardiol. 2011 Mar 15;107(6):817-20.e1. 497. Yang X, So WY, Kong AP, et al. doi: 10.1016/j.amjcard.2010.11.004. PMID: Development and validation of stroke risk 21247534. Exclusion Code: X5. equation for Chinese patients 504. Zhang ZM, Prineas RJ, Soliman EZ, et al. with type 2 diabetes: the Hong Kong Prognostic significance of serial Q/ST-T Diabetes Registry. Diabetes Care. 2007 changes by the Minnesota Code and Jan;30(1):65-70. doi: 10.2337/dc06-1273. Novacode in the Atherosclerosis Risk in PMID: 17192335. Exclusion Code: X3. Communities (ARIC) study. Eur J Prev 498. Yang X, So WY, Kong AP, et al. Cardiol. 2012 Dec;19(6):1430-6. doi: Development and validation of a total 10.1177/1741826711426091. PMID: coronary heart disease risk score in type 2 21997257. Exclusion Code: X5. diabetes mellitus. Am J Cardiol. 2008 Mar 505. Zhang ZM, Rautaharju PM, Prineas RJ, et 1;101(5):596-601. doi: al. Bundle branch blocks and the risk of 10.1016/j.amjcard.2007.10.019. PMID: mortality in the Atherosclerosis Risk in 18308005. Exclusion Code: X3. Communities study. J Cardiovasc Med 499. Yoo WS, Kim HJ, Kim D, et al. Early (Hagerstown). 2016 Jun;17(6):411-7. doi: detection of asymptomatic coronary artery 10.2459/jcm.0000000000000235. PMID: disease in patients with type 2 diabetes 25575277. Exclusion Code: X5. mellitus. Korean J Intern Med. 2009 506. Zhang ZM, Rautaharju PM, Prineas RJ, et Sep;24(3):183-9. doi: al. Race and Sex Differences in the 10.3904/kjim.2009.24.3.183. PMID: Incidence and Prognostic Significance of 19721853. Exclusion Code: X5. Silent Myocardial Infarction in the 500. Zhang Y, Agnoletti D, Iaria P, et al. Gender Atherosclerosis Risk in Communities difference in cardiovascular risk factors in (ARIC) Study. Circulation. 2016 May the elderly with cardiovascular disease in the 31;133(22):2141-8. doi: last stage of lifespan: the PROTEGER 10.1161/circulationaha.115.021177. PMID: study. Int J Cardiol. 2012 Feb 27185168. Exclusion Code: X5. 23;155(1):144-8. doi: 507. Zulqarnain MA, Qureshi WT, O'Neal WT, 10.1016/j.ijcard.2011.09.073. PMID: et al. Risk of Mortality Associated With QT 22000269. Exclusion Code: X2. and JT Intervals at Different Levels of QRS 501. Zhang Y, Post WS, Dalal D, et al. QT- Duration (from the Third National Health interval duration and mortality rate: results and Nutrition Examination Survey). Am J from the Third National Health and Cardiol. 2015 Jul 1;116(1):74-8. doi: Nutrition Examination Survey. Arch Intern 10.1016/j.amjcard.2015.03.038. PMID: Med. 2011 Oct 24;171(19):1727-33. doi: 25929581. Exclusion Code: X5. 10.1001/archinternmed.2011.433. PMID: 22025428. Exclusion Code: X5.

Screening for CVD Ri sk Wi th ECG 114 RTI–UNC EPC Appendix D Ta ble 1. Quality Assessment of Randomized, Controlled Trials (KQs 1 and 3): Part 1

Did the study Did the study have have cross- differential What w as the overs or What w as the attrition or overall First Author, Was Was allocation Were groups reported contamination What w as the differential high attrition Year randomization concealment similar at intervention raising concern overall attrition? attrition? raising concern Trial Nam e adequate? adequate? baseline? fidelity? for bias? No. (%) No. (%) for bias? Lievre, 201147 Yes Yes Yes NR NR 7 (1.1%) mortality 16 (1.2) No DY NA MIT 16 (2.5%) primary composite outcome Turrini, 201548 Yes Yes Yes 17/20 (85%) w ith Yes, 44 (17%) in NR NR Unclear and Turrini, positive exercise the no screening 200949 test underw ent group had non- DA DDY -D angiography protocol exercise testing (but unclear if those w ere indicated because of incident s y mptoms ) Abbreviations: DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; KQ=key question; No, number; NR=not reported.

Screening for CVD Ri sk Wi th ECG 115 RTI–UNC EPC Appendix D Ta ble 2. Quality Assessment of Randomized, Controlled Trials (KQs 1 and 3): Part 2

Was the duration of What w as the Were outcome Were followup method used Did the study First Author, measurements Were Were outcome adequate to to handle use acceptable Year equal, valid patients providers assessors assess the missing statistical Quality Trial Nam e and reliable? masked? masked? masked? outcome? data? methods? Rating Comments Lievre, 201147 Yes No No Yes Unclear None Yes Fair Trial stopped early because of trouble DY NA MIT (3.5 years) (complete recruiting (randomized 631 of the case analysis) planned 3,000). Not clear that 3.5 years of follow up is sufficient. Applicability: high risk population of diabetics w ith 2 risk factors (urinary albumin excretion above a threshold, hypertension, hyperlipidemia, PAD, history of TIA, tobacco consumption, and family history of premature CVD); Table 1 show s 14% w ith PA D, 4–5% w ith history of TIA ; patients referred to a diabetes specialist in a hospital. Study left subsequent investigations after stress test to judgment of cardiologists (no protocol; pragmatic approach for decisions about e.g., angiography or not, various treatments). Turrini, Yes No No NR Unclear NR Yes Fair Study did not reach sample size goal 201548 and (3.6 years) Complete (aimed for 364 per group and got about Turrini, case 260); not clear that 3.6 years of follow up 200949 is sufficient; amount of missing data NR DA DDY -D (flow diagram may indicate no missing data though); masking of outcome assessors NR. Applicability: setting w as 2 diabetes outpatient clinics, and participants had to have a normal ECG to get into the study. Abbreviations: CVD=cardiovascular disease; DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; ECG=electrocardiogram; KQ=key quest ion; NR=not reported; PAD=peripheral artery disease; TIA=transient ischemic attack.

Screening for CVD Ri sk Wi th ECG 116 RTI–UNC EPC Appendix D Ta ble 3. Quality Assessment of Randomized, Controlled Trials: Additional Questions for Studies Reporting Harms (KQ 3)

Were Were Was duration of ascertainment ascertainment follow up Were harms techniques for techniques for adequate for First Author, Year prespecified harms adequately harms equal, valid, harms Quality Trial Nam e and defined? described? and reliable? assessment? Rating Comments Turrini, 201548 and Yes for post- No Unclear Yes Fair (for MI) Study reports that one Turrini, 200949 procedure MI. but poor for patient had an MI 3 days DA DDY -D No for other other after a revascularization potential harms. harms procedure and that there w ere no other harms/events for those w ho underw ent revascularization; limited assessment of harms and no mention of, for example, postintervention hematomas or infections. Other than MI, no methods on how harms w ere defined or measured, if they w ere measured at all. Abbreviations: DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; KQ=key question; MI=myocardial infarction.

Screening for CVD Ri sk Wi th ECG 117 RTI–UNC EPC Appendix D Ta ble 4. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 1

Is a valid and reliable definition and Does study method for sample If Is a valid and Were the measurement adequately Was there participants reliable Was the same outcomes and capture the selective are from a definition and outcome assessed classification of population of inclusion of treatment Enrolled Were method for definition (and without candidate interest participants RCT, is consecutive selection measurement method for knowledge of predictor(s) (participant in the model treatment patients or a criteria of the measurement) the candidate (ECG and risk First Author, eligibility and based on data accounted random clearly outcomes used in all predictors (i.e., factors) Year recruitment)? availability? for? sample? described? reported? patients? blinded)? reported? Aktas, 200459 Yes, no CVD and NR NA Yes Yes Yes Yes NR Yes asymptomatic (but executive physical participants may not be representative) Auer, 201255 Yes Yes (but little NA Yes Yes Yes Yes Yes Yes missing data) Badheka, Somew hat Yes NA Complex, Yes Yes Yes Unknow n Yes 201353 unclear; NHANES nonrandom, III s ample; multistage symptoms not stratified assessed; just sample design under 10% w ith (NHA NES III) CA D Badheka, Yes Yes for NA Complex, Yes Yes Yes Unknow n Yes 201354 discrimination nonrandom and multistage calibration. No stratified f or NRI sample design calculations; those excluded 5% w ith missing data).

Screening for CVD Ri sk Wi th ECG 118 RTI–UNC EPC Appendix D Ta ble 4. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 1

Is a valid and reliable definition and Does study method for sample If Is a valid and Were the measurement adequately Was there participants reliable Was the same outcomes and capture the selective are from a definition and outcome assessed classification of population of inclusion of treatment Enrolled Were method for definition (and without candidate interest participants RCT, is consecutive selection measurement method for knowledge of predictor(s) (participant in the model treatment patients or a criteria of the measurement) the candidate (ECG and risk First Author, eligibility and based on data accounted random clearly outcomes used in all predictors (i.e., factors) Year recruitment)? availability? for? sample? described? reported? patients? blinded)? reported? Chang, 201545 Uncertain, No (although NA Unknow n No, “clinically Yes Yes Yes Yes for most risk possibly not, see unclear if the indicated factors and comments 4% w ithout reasons” for ECG/ETT. No, adequate CACS and cholesterol and follow up w ere stress blood pressure included in SPECT not w ere not analyses) defined available, and except for conservative the 16% w ith values w ere atypical imputed based chest pain on history of hypertension or hyperlipidemia. Cournot, Yes, regarding Yes NA Yes Yes Uncertain Yes Unknow n Yes 200660 asymptomatic (questionnaire status, but they or phone call w ere self-referred to patients and or referred by physicians for PCPs and initial cardiologists to a ascertainment) preventive cardiology unit Cournot, Yes, regarding Yes (but little NA No (referrals Yes Uncertain Yes Unknow n Yes 200956 asymptomatic missing data) and from (questionnaire status, but they media) or phone call w ere self-referred to patients and (20%) or referred physicians) by PCPs (27%) or other providers to a preventive cardiology unit

Screening for CVD Ri sk Wi th ECG 119 RTI–UNC EPC Appendix D Ta ble 4. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 1

Is a valid and reliable definition and Does study method for sample If Is a valid and Were the measurement adequately Was there participants reliable Was the same outcomes and capture the selective are from a definition and outcome assessed classification of population of inclusion of treatment Enrolled Were method for definition (and without candidate interest participants RCT, is consecutive selection measurement method for knowledge of predictor(s) (participant in the model treatment patients or a criteria of the measurement) the candidate (ECG and risk First Author, eligibility and based on data accounted random clearly outcomes used in all predictors (i.e., factors) Year recruitment)? availability? for? sample? described? reported? patients? blinded)? reported? Denes, 200758 Yes (but only Yes (<10% of Yes No Yes Y es f or CV D; Yes Unknow n Yes females w ith intact the larger unclear uterus in WHI) sample had w hether measured inclusion of cholesterol silent MI from data allow ing ECGs w ithin calculation of the CHD FRS) composite outcome is valid and reliable Erikssen, Yes Unknow n (full NA Yes Yes Yes Yes Unknow n Yes 200462 sample, survey 1), yes (survey 2) Folsom, Yes Yes NA Yes Yes Yes Yes Unknow n Yes 200363 Ishikaw a, Yes Yes NA Yes Yes Yes Yes Unknow n Yes 201550 Jorgensen, Yes Yes NA Yes Yes Yes Yes Unknow n Yes 201452 Shah, 201657 Probably; Yes NA Complex, Yes Yes Yes Unknow n Yes NHA NES I and III (3,640/4,192 nonrandom, samples; w ith ECG data multistage, symptoms not for derivation stratified assessed, but self- cohort, sample reported CVD NHA NES I; (NHA NES I excluded 6,329/6,927 and III) w ith ECG for validation c ohort (III)

Screening for CVD Ri sk Wi th ECG 120 RTI–UNC EPC Appendix D Ta ble 4. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 1

Is a valid and reliable definition and Does study method for sample If Is a valid and Were the measurement adequately Was there participants reliable Was the same outcomes and capture the selective are from a definition and outcome assessed classification of population of inclusion of treatment Enrolled Were method for definition (and without candidate interest participants RCT, is consecutive selection measurement method for knowledge of predictor(s) (participant in the model treatment patients or a criteria of the measurement) the candidate (ECG and risk First Author, eligibility and based on data accounted random clearly outcomes used in all predictors (i.e., factors) Year recruitment)? availability? for? sample? described? reported? patients? blinded)? reported? Strom Moller, Population-based, Yes NA Yes Yes Yes Yes Unknow n Yes 200761 but uncertain how many people w ith prior ASCVD or symptoms w ere included Tereshchenko, Yes, but uncertain Yes NA Yes Yes Yes Yes Unknow n Yes 201451 how many had CVD at baseline depending on overlap or lack thereof for those w ith history of CA D, MI, HF, and stroke (could be as low as 5% or as high as 14%) Abbreviations: ASCVD=arteriosclerotic cardiovascular disease; CACS=coronary artery calcium score; CAD=coronary artery disease; CVD=cardiovascular disease; ECG=electrocardiogram; ETT=exercise treadmill test; FRS=Framingham Risk Score; KQ=key question; HF=heart failure; NA=not applicable; NHANES=National Health and Nutrition Examination Survey; NRI=Net reclassification index or improvement; PCP=primary care physician; RCT=randomized, controlled trial; SP ECT =single-photon emission computed tomography; WHI=Women’s Health Initiative.

Screening for CVD Ri sk Wi th ECG 121 RTI–UNC EPC Appendix D Ta ble 5. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 2

Were Were a priori predictors Were multiple cut points assessed measures of Were both used for Was a bias- blinded Number of (%) performance calibration and classification cor r e cte d NRI for the Participants Did the used (e.g., global discrimination measures used? This outcome How was the With Missing study have fit, measures (e.g., applies only to and for predictor of Data (Include high attrition discrimination, reported? Were sensitivity, studies each other interest (ECG) Predictors raising How was calibration, net confidence specificity, presenting NRI First Author, (if handled in the and concern for missing data reclassifica- intervals predictive for a specific Year relevant)? modeling? Outcomes) bias? handled? tion)? reported? values, NRI)? risk strata? Aktas, 200459 NR Categorized NR NR NR No Discrimination NA NA only reported, no CIs Auer, 201255 Yes Categorized For main No Complete- Yes Calibration: p- Yes Yes analyses, case analysis values excluded those w ith missing Discrimination data on w ith CIs traditional risk factors (n=41, <2%); for secondary analyses of people w ith follow up ECGs at 4 years, excluded 424 (24%) w ith missing follow up ECGs Badheka, Unknow n Categorized 62 (<1%) No Complete- Yes Both reported, Yes NA 201353 missing case analysis w ith CIs only for mortality (n=4) (related to ST- discrimination or ST-T data T and (also given for (n=58) mortality; NRI) some multiple imputation used for other variables) Badheka, Unknow n Categorized 5% No Multiple Yes Both reported, Yes NA 201354 ( ECG (breakdow n imputation CI only reported abnormalities NR) for absent vs. discrimination present)

Screening for CVD Ri sk Wi th ECG 122 RTI–UNC EPC Appendix D Ta ble 5. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 2

Were Were a priori predictors Were multiple cut points assessed measures of Were both used for Was a bias- blinded Number of (%) performance calibration and classification cor r e cte d NRI for the Participants Did the used (e.g., global discrimination measures used? This outcome How was the With Missing study have fit, measures (e.g., applies only to and for predictor of Data (Include high attrition discrimination, reported? Were sensitivity, studies each other interest (ECG) Predictors raising How was calibration, net confidence specificity, presenting NRI First Author, (if handled in the and concern for missing data reclassifica- intervals predictive for a specific Year relevant)? modeling? Outcomes) bias? handled? tion)? reported? values, NRI)? risk strata? Chang, 201545 Yes Categorized 4% missing No, not for Imputation for Yes Both reported, Yes Unknow n (ETT show ing follow up for lost to cholesterol and no CIs (p-values presence or outcomes; follow up (but blood pressure given) absence of 100% missing concern for based on ischemia; i.e., cholesterol and missing data history of high or low risk) blood pressure on some diagnosis of numbers; NR predictors) hypertension for other or predictors hyperlipidemia; unclear for missing outcome data (likely complete case analysis) Cournot, 200660 Unknow n Categorized Lost to No Complete- No Calibration Cut point used NA follow up n=138 case analysis reported w ithout and reported (11%); missing CIs ; but not a priori predictors discrimination (w as based on unknow n not reported median FRS score) Cournot, 200956 Unknow n Categorized 87 (3.4%) No Complete- Yes Both reported, Unclear (cut NA case analysis no CIs point reported, but unclear if a priori decision) Denes, 200758 Unknow n Categorized 91% Unknow n (for Complete- No (only Discrimination NA NA (normal, minor (breakdow n the sample of case analysis discrimination) only reported, abnormalities, NR) for the interest to us) w ith CIs or major only eligible abnormalities) results (Figure 3 and the related text)

Screening for CVD Ri sk Wi th ECG 123 RTI–UNC EPC Appendix D Ta ble 5. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 2

Were Were a priori predictors Were multiple cut points assessed measures of Were both used for Was a bias- blinded Number of (%) performance calibration and classification cor r e cte d NRI for the Participants Did the used (e.g., global discrimination measures used? This outcome How was the With Missing study have fit, measures (e.g., applies only to and for predictor of Data (Include high attrition discrimination, reported? Were sensitivity, studies each other interest (ECG) Predictors raising How was calibration, net confidence specificity, presenting NRI First Author, (if handled in the and concern for missing data reclassifica- intervals predictive for a specific Year relevant)? modeling? Outcomes) bias? handled? tion)? reported? values, NRI)? risk strata? Erikssen, Unknow n Categorized Not reported Not reported Not reported No Calibration Yes NA 200462 (survey 1); for reported w ithout survey 2 CIs (number of 586/2,014 predicted and (29%) not observed included events); discrimination not reported Folsom, 200363 Unknow n Categorized 11% No Complete- Yes Both reported, Unknow n NA case analysis no CIs Ishikaw a, Yes Categorized Overall <14%; No Complete- Yes Neither reported Yes NA 201550 and continuous No ECG case analysis (both w ere (n=1,285), done in Incomplete separate data (n=5) models) No follow up data (n=84) Jorgensen, Unknow n Categorized NR Unknow n Other Yes Discrimination No, NRI c ut NA 201452 (complete- only reported, points for risk case analysis w ith CIs categories except that w ere based on missing HDLs the data imputed by setting to mean of remaining participants) Shah, 201657 Unknow n Some variables 641 (6%) No Complete- Yes Both reported, Yes NA categorized missing ECG case analysis w ith CIs for and some data discrimination continuous NR f or only outcomes

Screening for CVD Ri sk Wi th ECG 124 RTI–UNC EPC Appendix D Ta ble 5. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 2

Were Were a priori predictors Were multiple cut points assessed measures of Were both used for Was a bias- blinded Number of (%) performance calibration and classification cor r e cte d NRI for the Participants Did the used (e.g., global discrimination measures used? This outcome How was the With Missing study have fit, measures (e.g., applies only to and for predictor of Data (Include high attrition discrimination, reported? Were sensitivity, studies each other interest (ECG) Predictors raising How was calibration, net confidence specificity, presenting NRI First Author, (if handled in the and concern for missing data reclassifica- intervals predictive for a specific Year relevant)? modeling? Outcomes) bias? handled? tion)? reported? values, NRI)? risk strata? Strom Moller, Yes for the Categorized For the 70- Yes Complete case No Discrimination NA NA 200761 outcome, year-old only reported, NR f or cohort, no CIs (data each other 1,139/2,239 only for the (51%) had subgroup w ho ECGs and had follow up w ere included ECGs at age (1,139/1,681, 70) 68%, of those w ho had been invited) Tereshchenko, Unknow n Categorized 2.6% No Complete- Yes Neither reported Yes Unknow n 201451 case analysis for eligible comparisons Abbreviations: CI=confidence interval, ECG=electrocardiogram; ETT=exercise treadmill test; FRS=Framingham Risk Score; HDL=high-density lipoprotein cholesterol; KQ=key question; NA=not applicable; NR=not reported; NRI=net reclassification improvement.

Screening for CVD Ri sk Wi th ECG 125 RTI–UNC EPC Appendix D Ta ble 6. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 3

Appendix D Ta bl e 1. Qual i t y R at in g s fo r Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 3

Method used for Was the FRS or If net reclassification testing model In w hat w ay w as PCE recalibrated was assessed, w ere performance: the population a in the population appropriate clinical development dataset separate external before ECG was First Author, thresholds used to only or separate validation from added to the Year reclassify risk? external validation? the FRS or PCE? model? Quality Comments Aktas, 200459 NA Developmental dataset Entirely different NA (did not use Fair Provides limited information relevant to our only w ith respect to population (but NA FRS or PCS f or questions (just some discrimination statistics SCORE + ex erc is e ECG because did not the parts eligible w ithout CIs); masking NR, amount and handling models (for models w ith use FRS or PCS; for our evaluation; of missing data NR; mean follow up of only 8 SCORE alone, could be used SCORE) used SCORE) years (less than desired for 10-year risk considered external prediction). Population referred for executive validation) physical. Auer, 201255 Yes Developmental dataset NA Yes Good Applicability to older patients, ages 70–79 only (no split of data) years at baseline; good internal validity but developmental dataset and no validation set; did not use FRS because it has not been validated in adults over age 75, but adjusted for traditional risk factors included in FRS and diabetes; used 7.5% to 15% risk thresholds over 7.5 years (attempting to correspond w ith 10–20% 10-year risk); mean follow up w as 6.4 years Badheka, Used FRS 5%–20% Developmental dataset Entirely different Unknow n (but Fair Unclear proportion of population w ith 201353 for intermediate risk population seems that it w as symptoms; masking not reported; used 5%– category (not 7.5% or (NHA NES III) not) 20% for intermediate risk category 10%) Badheka, Yes Developmental dataset Entirely different Unknow n Fair 201354 only (no split of data) population (used NHA NES)

Screening for CVD Ri sk Wi th ECG 126 RTI–UNC EPC Appendix D Ta ble 6. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 3

Method used for Was the FRS or If net reclassification testing model In w hat w ay w as PCE recalibrated was assessed, w ere performance: the population a in the population appropriate clinical development dataset separate external before ECG was First Author, thresholds used to only or separate validation from added to the Year reclassify risk? external validation? the FRS or PCE? model? Quality Comments Chang, 201545 Used <6% vs. 6–20% Developmental dataset Entirely different Yes (unable to use Fair Moderate concern that the population may vs. >20% (not 7.5% or only population (but NA original FRS have limited applicability to our question; men 10%) because did not coefficients and w omen w ith no history of CAD w ho had use original FRS because of coronary artery calcium and stress SPECT coefficients) missing blood performed for “clinically indicated reasons”; pressure and 16.5% w ith atypical chest pain and unclear how cholesterol values) many had other symptoms; risk of misclassification due to imputing unavailable/ missing data for all cholesterol and blood pressure measurements based on prior diagnoses of hypertension and hyperlipidemia (measurements w ere not available so they calculated FRS using conservative imputations; page 135); did not use current clinical thresholds for reclassification; follow up w as 6.9 years; this is a derivation study w ithout external validation. Calibration NR, p-values for discrimination. Cournot, NA Developmental dataset Entirely different No Fair Masking of outcome assessors and assessors 200660 w ith respect to FRS + population of relevant exposures NR; uncertain validity of exercise ECG models; outcome assessment procedures (relied for models w ith FRS primarily on questionnaire and phone calls for alone, could be initial ascertainment); calibration reported, but considered external w ithout CIs, and reclassification and validation of FRS discrimination NR; duration of follow up mean 6 years. Perhaps limited applicability of the selected population that included many referrals.

Screening for CVD Ri sk Wi th ECG 127 RTI–UNC EPC Appendix D Ta ble 6. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 3

Method used for Was the FRS or If net reclassification testing model In w hat w ay w as PCE recalibrated was assessed, w ere performance: the population a in the population appropriate clinical development dataset separate external before ECG was First Author, thresholds used to only or separate validation from added to the Year reclassify risk? external validation? the FRS or PCE? model? Quality Comments Cournot, NA Developmental dataset Entirely different No Fair Masking of outcome assessors and assessors 200956 w ith respect to FRS + population of relevant exposures NR; uncertain validity of exercise ECG models; outcome assessment procedures (relied for models w ith FRS primarily on questionnaire and phone calls for alone, could be initial ascertainment); some measures of considered external discrimination and calibration reported, but validation of FRS without CIs, and reclassification NR; duration of follow up median 6 years. The relevant models also included femoral bruit in addition to adding exercise ECG. Perhaps limited applicability of the selected population that included many referrals Denes, 200758 NA Developmental dataset Entirely different No Fair High proportion of missing data for the only (no split of data) population (used analyses eligible for our questions; less than WHI) 10% of the sample w as included in the analyses eligible for our questions (due to lack of measured cholesterol information for most participants); complete-case analysis; only reports C-statistic Erikssen, NA Developmental dataset NA NA Fair Masking of outcome assessors NR, lack of 200462 only multiple measures (calibration only; did not report discrimination or reclassification), missing data NR Folsom, 200363 NA Developmental dataset Entirely different NA (did not use Fair Masking NR; developmental set only; unclear if only population (ARIC) FRS or PCE) cut points for LVH w ere defined a priori; reclassification NR Ishikaw a, No, used <2.5%, 2.5– Development dataset NA NA (did not use Fair Base model included alcohol use and heart rate 201550 5%, >5% only FRS or PCE; used in addition to traditional risk factors; unclear model w ith w hy these particular clinical thresholds w ere traditional risk chosen (<2.5%, 2.5–5%, >5%). Unknow n if factors plus heart predictors w ere assessed blinded for outcomes; rate and alcohol ECG QTc (predictor) determined by hand. use as base model)

Screening for CVD Ri sk Wi th ECG 128 RTI–UNC EPC Appendix D Ta ble 6. Quality Ratings for Studies Reporting Reclassification, Calibration, and Discrimination (KQ 2): Part 3

Method used for Was the FRS or If net reclassification testing model In w hat w ay w as PCE recalibrated was assessed, w ere performance: the population a in the population appropriate clinical development dataset separate external before ECG was First Author, thresholds used to only or separate validation from added to the Year reclassify risk? external validation? the FRS or PCE? model? Quality Comments Jorgensen, No, used cut points Developmental dataset Entirely different NA, used different Fair NRI cut points for risk categories w ere based 201452 determined by the only population (but NA base model w ith on the data; calibration NR; unknow n masking; data (low risk: <8.9%; because did not conventional risk amount of missing ECG data NR intermediate use FRS or PCE) factors risk: 8.9%–14.9; high risk: >14.9%).a Shah, 201657 Used 1%, 5%, and Developmental dataset Entirely different Yes and no Fair Masking NR; some uncertainty about symptom 10% based on (NHA NES I) and population (appears that they status of participants recommendation by validation (NHANES III) ran analyses both the European Society w ays) of Cardiology for lipid management Strom Moller, NA Developmental dataset NA NA Poor Very high attrition. Study started at age 50. The 200761 only eligible analysis/outcome is the part w here they report discrimination (ROC curve) for FRS (traditional risk factors) vs. FRS + ECG indicating ischemia for a subgroup of the cohort w ho had repeated ECGs (20 years later) at age 70 (n=1,139 of the 2.239, 51%); they had 12- year follow up after age 70. No similar analyses provided for the full cohort of 2,239 participants. Also, unclear selection criteria; unclear w hat proportion of the population had prior CVD or symptoms at baseline for the 50-year-old initial cohort or for the 70-year-old subgroup. Tereshchenko, Used <5%, 5–<20% Developmental dataset Entirely different Yes (FRSs w ere Fair Did not report calibration or discrimination for 201451 and ≥20% (not 10% or only population (used not directly used eligible comparison (ECG finding+traditional 7.5%) A RIC) due to possible risk factors vs. traditional risk factors alone); issues of the unclear masking; did not use 10% or 7.5% applicability to threshold in classifications of risk groups different ethnic groups; the authors adjusted for race) a Supplement table 2 reported different risk categories for categorical NRI – low risk < 23.8%, intermediate 23.8–35%, high >35% Abbreviations: ARIC=Atherosclerosis Risk in Communities study; CAD=coronary artery disease; CI=confidence interval; CVD=cardiovascular disease; ECG=electrocardiogram; KQ=key question; LVH=left ventricular hypertrophy; NA=not applicable; NHANES=National Health and Nutrition Examination Survey; NR=not reported; PCE=pooled cohort equation; RCT=randomized, controlled trial; ROC=receiver operating characteristic; SCORE=Systematic COronary Risk Evaluation; SP ECT =single-photon emission computed tomography; WHI=Women’s Health Initiative.

Screening for CVD Ri sk Wi th ECG 129 RTI–UNC EPC Appendix E Ta ble 1. Results of Included Randomized, Controlled Trials Reporting Health Outcomes (KQ 1)

All-Cause Mortality CV Mortality MI He ar t Failure Stroke Othe r CV Eve n t s Composite CV Outcome Fir s t Author, G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) Year G1 (N) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) Trial Name G2 (N) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) Lievre, Screened 15 (4.7) NR 4 (1.3) Hospitalized cardiac 9 (2.8) Revascularization Main endpointa 201147 (316) 13 (4.1) 8 (2.5) failure: 4 (1.3) 18 (5.7) 28 (8.9) DY NA MIT Not NR, NS NR, NS 5 (1.6) NR, NS 21 (6.7) 26 (8.3) screened 4 (1.3) p=0.61 1.00 (0.59 to 1.71) (315) NR, NS Coronary eventsb 13 (4.1) 15 (4.8) 0.77 (0.37 to 1.63) Turrini et al, Screened NR by All Non-fatal MI: All NR by NR Cardiac events (primary 201548 (262) group. 1 (0.4) All 2 (0.8) group. Total outcome; composite of DA DDY -D Not Total of 19 5 (1.9) 11 (4.2) 7 (2.7) of 7 strokes nonfatal MI or cardiac death) screened deaths 0.197 (0.023 to 12 (4.7) 0.273 (0.57 to 1.314) All (258) reported (6 1.683) 0.908 (0.400 to 2.057) Female 12 (4.6) cardiac and Female Female 0/53 (0) 14 (5.4) 13 non- 0/53 (0) 5/53 (9.4) 3/51 (5.9) 0.849 (0.393 to 1.837) cardiac) 3/51 (5.9) 1/51 (2) NA, p 0.065 Female NA , p=0.077 4.916 (0.573 to 5/53 (9.4) 42.142) 3/51 (5.9) 1.653 (0.395 to 6.928)

Screening for CVD Ri sk Wi th ECG 130 RTI–UNC EPC Appendix E Ta ble 1. Results of Included Randomized, Controlled Trials Reporting Health Outcomes (KQ 1)

All-Cause Mortality CV Mortality MI He ar t Failure Stroke Othe r CV Eve n t s Composite CV Outcome Fir s t Author, G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) Year G1 (N) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) Trial Name G2 (N) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) Turrini et al, Male Male Male Male 201548 1/209 (0.5) 6/209 (2.9) 2/209 (1) 7/209 (3.3) DA DDY -D 2/207 (1) 11/207 (5.3) 4/207 (1.9) 11/207 (5.3) (continued) 0.497 (0.045 to 0.535 (0.198 to 1.446) 0.485 (0.89 to 2.647) 0.625 (0.242 to 1.614) 5.483) >60 years >60 years >60 years >60 years 7/182 (3.8) 2/182 (1.1) 7/182 (3.8) 0/182 (0) 8/181 (4.4)0.859 4/181 (2.2) 10/181 (5.5) 4/181 (2.2) (0.311 to 2.370) 0.476 (0.087 to 0.687 (0.261 to 1.805) NA , p=0.044 <60 years 2.599) <60 years <60 years 4/80 (5) <60 years 5/80 (6.3) 1/80 (1.3) 4/77 (5.2) 0/80 (0) 4/77 (5.2) 1/77 (1.3) 0.999 (0.249 to 3.981) 3/77 (3.9) 1.264 (0.339 to 4.707) 0.993 (0.062 to CV risk>20 NA, p 0.08 CV risk>20 15.87) 4/115 (3.5) CV risk>20 5/115 (4.3) CV risk>20 6/112 (5.4) 2/115 (1.7) 6/112 (5.4) 1/115 (0.9) 0.656 (0.185 to 2.325) 2/112 (1.8) 0.822 (0.251 to 2.692) 2/112 (1.8) CV risk<20 0.961 (0.135 to CV risk<20 0.494 (0.045 to 7/147 (4.8) 6.822) 7/147 (4.8) 5.443) 6/146 (4.1) CV risk<20 8/146 (5.5) CV risk<20 1.176 (0.393 to 3.505) 0/147 (0) 0.879 (0.318 to 2.426) 0/147 (0) 5/146 (3.4) 3/146 (2.1) NA , p=0.022 NA , p=0.08 a Composite endpoint of death from all causes, nonfatal MI, nonfatal stroke, or heart failure requiring hospitalization or emergency service intervention. b Defined as fatal or nonfatal MI, hospitalized unstable angina, or heart failure requiring hospitalization or emergency service intervention.

Abbreviations: CI=confidence interval; CV=cardiovascular; DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; KQ=key quest ion; G=group; HR=hazard ratio; MI=myocardial infarction; N=sample size; NA=not applicable; NR=not reported; NS=not significant.

Screening for CVD Ri sk Wi th ECG 131 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Auer, 201255 CHD Total: 351 (16.0) Base model: 1. CVRF model Harrell C Index Hosmer-Lemeshow CV RF + ECG v s . CV RF Fair eventsa CV RF (used FRS (95% CI) chi-square Overall sample CHD deaths: 96 Harrell C index variables)b CVRF model: 0.58 CVRF model: 67.6 NRI: 7.4% (3.1%–19.0%) (4.4) Calibration 2. CV RF (0.53–0.62) CV RF + ECG: 87.9 Ev ent NRI: -0.9% Acute MIs: 101 NRI + any ECG CV RF + ECG: 0.60 Likelihood ratio: Nonevent NRI: 8.3% (4.6) Adjusted abnormality (0.56–0.65) p≤0.00005 Adjusted clinical NRI 6.7% (95% CI, Hospitalizations for clinical NRI 3. FRS model 1.2% to 19.3%) angina or coronary IDI 4. FRS + any Goodness of fit p IDI: 0.99% (0.32%–2.15%) revascularization: Base model: ECG value For the Intermediate Risk Category: 154 (7.0) FRS abnormality CVRF model: 0.03 NRI 13.6% NRI CVRF model + IDI ECG: 0.01 FRS + ECG v s . FRS NRI 5.7% (-0.4%–11.8%) IDI: 1.03% (0.56%–1.50%)

Reclassification w ith Addition of ECG For those w ho experience a CHD event <7.5%, no change: 4 <7.5% to 7.5-<15%: 2 <7.5% to >15%: 0 Total: 6

7.5-<15.0% to <7.5%: 7 7.5-<15.0%, no change: 91 7.5-<15% to >15%: 27 Total: 125

>15.0% to <7.5%: 0 >15.0% to 7.5-<15.0%: 25 >15%, no change: 195 Total: 220

For those w ho do not experience a CHD ev ent <7.5%, no change: 74 <7.5% to 7.5-<15%: 17 <7.5% to >15%: 0

Total: 91 7.5-<15.0% to <7.5%: 129 7.5-<15.0%, no change: 678 7.5- <15% to >15%: 149 Total: 956

Screening for CVD Ri sk Wi th ECG 132 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Auer, 201255 >15.0% to <7.5%: 0 Fair >15.0% to 7.5-<15.0%: 189 (continued) >15%, no change: 605 Total: 794 Badheka, CV mortality CV mortality: 739 A UC Model A: FRSc C-statistic Hosmer-Lemeshow Overall NRI: 3.6%, p=0.0001; 201354 (12.3) IDI Model A: 0.851 chi-square NRI for Those w ith Events: 3.0% Fair All-cause Calibration Model B: FRS (0.836–0.865) Model A: 15.14 p=0.03 mortality All-cause mortality: NRI + ECG Model B: 0.852 p=0.05 NRI for Those w ithout Events: NRI: 1,824 (30.3) abnormalities (0.838–0.866) Model B: 10.98 0.6% p=0.11 p=0.05 p=0.2 Absolute IDI: 0.0001, p≤0.001 For the Intermediate Risk Category: Goodness of fit NRI 13.2% Likelihood ratio test: p=0.001 Reclassification w ith Addition of ECG Abnormalities Bayesian For Those w ith Events information criterion <5%, no change: 92 Model A: 3360.54 <5% to 5-10%: 11 Model B: 3358.28 <5% to 10–20%: 0 <5% to >20%: 0 Total: 103

5-10% to <5%: 4 5-10%, no change: 98 5-10% to 10–20%: 26 5-10% to >20%: 0 Total: 128

10-20% to <5%: 0 10-20% to 5–10%: 14 10-20%, no change: 205 10-20% to >20%: 21 Total: 240

>20% to <5%: 0 >20% to 5-10%: 0 >20% to 10–20%: 19 >20%, no change: 208 Total: 227

Screening for CVD Ri sk Wi th ECG 133 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Badheka, Risk Reclassified Higher from Each 201354 Category Fair For Those w ith Events (continued) <5%: 11 (10.7%) 5% to <10%: 26 (20.3%) 10% to <20%: 21 (8.8%) >20%: NA Total: 58 (8.3%)

Risk Reclassified Low er from Each Category For Those w ith Events <5%: NA 5% to <10%: 4 (3%) 10% to <20%: 14 (5.8%) ≥20%: 19 (8.4%) Total: 37 (5.3%)

For Those w ith no Events <5%, no change: 3209 <5% to 5–10%: 79 <5% to 10–20%: 0 <5% to >20%: 0 Total: 3,288

5-10% to <5%: 104 5-10%, no change: 702 5-10% to 10–20%: 53 5-10% to >20%: 0 Total: 859

10-20% to <5%: 0 10-20% to 5–10%: 65 10-20%, no change: 512 10-20% to >20%: 37 Total: 614

>20% to <5%: 0 >20% to 5–10%: 0 >20% to 10–20%: 31 >20%, no change: 239 Total: 270

Screening for CVD Ri sk Wi th ECG 134 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Badheka, Risk Reclassified Higher 201354 No Events Fair <5%: 79 (2.4%) (continued) 5% to <10%: 53 (6.2%) 10% to <20%: 37 (6.0%) >20%: NA Total: 169 (3.4%)

Risk Reclassified Low er No Events <5%: NA 5% to <10%: 104 (12.1%) 10% to <20%: 65 (10.6%) ≥20%: 31 (11.5%) Total: 200 (4.0%)

Intermediate Risk Cohort: NRI: 13.24% 137 (2.4%) and 187 (3.3%) w ere reclassified to higher and low er risk groups, respectively

Screening for CVD Ri sk Wi th ECG 135 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Badheka, Cardiovas- 1,226 (15.5) C-statistic Model A: FRS C-statistic Hosmer-Lemeshow Overall NRI: 0.07 (0.05–0.09) p<0.01 201353 cular (CV) A UROC only Model A: 0.832 chi-square Reclassification of subjects w ith Fair mortality Calibration (0.822–0.841) Model A: 13.7 events: 2.7% p<0.01 Reclassification Model B: FRS Model B: 0.838 Model B: 16.7 Reclassification of subjects w ithout IDI plus the (0.828–0.848) events: 2.3% p<0.01 NRI variable T-w ave p<0.01 Goodness of fit Absolute IDI amplitude in Likelihood ratio test: 0.012 (0.009–0.015), p<0.01 lead aVR A UROC p<0.01 Relative IDI 0.11 Model A: 0.812 (0.800–0.824) Bayesian Participants w ith CVD disease events Model B: 0.820 information criterion <5%, no change: 53 (0.807–0.832) Model A: 5240.0 <5% to 5–<10%: 11 p<0.01 Model B: 5172.2 <5% to 10–<20%: 0 <5% to >20%: 0 Total: 64

5-<10% to <5%: 6 5-<10%, no change: 50 5-<10% to 10–<20%: 22 5-<10% to >20%: 0 Total: 78

10-<20% to <5%: 0 10-<20% to 5–<10%: 13 10-<20%, no change: 148 10-<20% to >20%: 60 Total: 221

>20% to <5%: 0 >20% to 5–<10%: 0 >20% to 10-<20%: 42 >20%, no change: 754 Total: 796

Screening for CVD Ri sk Wi th ECG 136 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Badheka, Participants w ith CVD disease events 201353 reclassified as higher risk from each Fair category (continued) <5%: 11 5%–<10%: 22 10%–<20%: 60 >20%: NA Total: 93

Participants w ith CVD disease events reclassified as low er risk from each category <5%: NA 5%–<10%: 6 10%–<20%: 13 >20%: 42 Total: 61

Participants w ithout CVD disease events <5%, no change: 2,426 <5% to 5–<10%: 135 <5% to 10–<20%: 0 <5% to >20%: 0 Total: 2,561

5–<10% to <5%: 221 5–<10%, no change: 764 5–<10% to 10–<20%: 158 5–<10% to >20%: 0 Total: 1,143

10–<20% to <5%: 0 10–<20% to 5-<10%: 216 10–<20%, no change: 898 10–<20% to >20%: 152 Total: 1,266

Screening for CVD Ri sk Wi th ECG 137 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Badheka, >20% to <5%: 0 201353 >20% to 5-<10%: 0 Fair >20% to 10–<20%: 158 (continued) >20%, no change: 12,22 Total: 1,380

Participants w ithout CVD disease events reclassified as higher risk from each category <5%: 135 5%–<10%: 158 10%–<20%: 152 >20%: NA Total: 445

Participants w ithout CVD disease events reclassified as low er risk from each category <5%: NA 5%–<10%: 221 10%–<20%: 216 >20%: 158 Total: 595

Intermediate risk cohort (5% to <20% risk): 60 (20%) of subjects w ith events and 219 (9.1%) of subjects w ithout events w ere reclassified appropriately to higher and low er risk categories, respectively Denes, 200758 Incident CHD events: 246 A UC FRS A UROC (95% CI) Likelihood ratio chi NR Fair coronary (1.7) Calibration/ FRS f or CHD: 0.69 square test heart overall FRS + ECG (0.61–0.86) FRS + ECG disease CVD events: 595 performance abnormality FRS + ECG abnormality (CHD) d (4) abnormality for CHD: p=0.004 CHD: 0.74 (0.66– CV D: p=0.02 Incident 0.90) cardiovas- FRS for CVD: 0.68 cular (0.62-0.77) disease FRS + ECG eventse abnormality for CVD: 0.70 (0.65– 0.79)

Screening for CVD Ri sk Wi th ECG 138 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Folsom, CHD ev ents f Total: 954 (6.8) A UROC Basic model A UROC NR NR 200363 Calibration Basic Model Fair Among 707 w omen Basic + LVH Women w ith w ith diabetes: 99 model diabetes: 0.711 (14.0) Women w ithout Among 6,526 diabetes: 0.777 w omen w ithout Men w ith diabetes: diabetes: 211 (3.2) 0.680 Among 566 men Men w ithout w ith diabetes: 129 diabetes: 0.679 (22.8) Among 4,946 men Basic + LVH Model w ithout diabetes: (yes/no) 515 (10.4) Women w ith diabetes: 0.709 Women w ithout diabetes: 0.777 Men w ith diabetes: 0.681 Men w ithout diabetes: 0.679

Screening for CVD Ri sk Wi th ECG 139 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Ishikaw a, Stroke Total: 375 (3.5) NRI Traditional NR NR IDI 201550 events Cerebral IDI cardiovascular The model w ith and w ithout QTc Fair hemorrhages: 85 risk factor Interval (as a continuous variable) (0.8) modelh (model) IDI= 0.292, p=0.80 Ischemic strokes: 242 (2.3) Model w ith The model w ith and w ithout ECG-LVH Subarachnoid ECG-LV H IDI= 0.004, p=0.75 hemorrhages: 47 w ithout QTc (0.4) interval The model w ith and w ithout ECG-LV H Unknow n cause: 1 and/or QTc interval Model w ith IDI=0.006, p=0.63 ECG-LVH w ith QTc interval NRI, Frequency (Percentage) The model w ith and w ithout QTc Model w ithout Interval (as a continuous variable) QTc interval (as Categorical NRI= 0.026, p<0.001 a continuous Event NRI 1.35% variable) Nonevent NRI 1.22%

Model w ith QTc Reclassification of subjects with stroke interval (as a events using the model wi th QTc continuous interval variable) <2.5% to <2.5%: 41 (91.1) <2.5% to 2.5–5.0%: 2 (3.4) Model w ithout <2.5% to >5.0%: 0 (0.0) ECG-LVH Total: 43

Model w ith 2.5–5.0% to <2.5%: 4 (8.9) ECG-LVH 2.5–5.0% to 2.5–5.0%: 51 (87.9) 2.5–5.0% to >5.0%: 11 (5.7) Model w ithout Total: 66 ECG-LVH or QTc interval >5.0% to <2.5%: 0 (0.0) >5.0% to 2.5–5.0%: 5 (8.6) Model w ith >5.0% to >5.0%: 183 (94.3) ECG-LVH and Total: 188 QTc interval

Screening for CVD Ri sk Wi th ECG 140 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Ishikaw a, Reclassification of subjects without 201550 stroke events using the model wi th Fair QTc interval (continued) <2.5% to <2.5%: 4,401 (95.5) <2.5% to 2.5–5.0%: 162 (9.2) <2.5% to >5.0%: 2 (0.1) Total: 4,565

2.5–5.0% to <2.5%: 205 (4.5) 2.5–5.0% to 2.5–5.0%: 1,384 (78.2) 2.5–5.0% to >5.0%: 165 (9.3) Total: 1,754

>5.0% to <2.5%: 0 (0.0) >5.0% to 2.5–5.0%: 224 (12.7) >5.0% to >5.0%: 1,613 (90.6) Total: 1,837

The model w ith and w ithout ECG-LVH Categorical NRI= 0.020, p<0.001 Event NRI 1.01% Nonevent NRI 1.01%

Reclassification of subjects with stroke events using the model with ECG-LVH

<2.5% to <2.5%: 41 (95.3) <2.5% to 2.5–5.0%: 2 (3.2) <2.5% to >5.0%: 0 (0.0) Total: 43 (14.5)

2.5–5.0% to <2.5%: 2 (4.7) 2.5–5.0% to 2.5–5.0%: 56 (88.9) 2.5–5.0% to >5.0%: 8 (4.2) Total: 66 (22.2)

Screening for CVD Ri sk Wi th ECG 141 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Ishikaw a, >5.0% to <2.5%: 0 (0.0) 201550 >5.0% to 2.5–5.0%: 5 (7.9) Fair >5.0% to >5.0%: 183 (95.8) (continued) Total: 188 (63.3)

Reclassification of subjects without stroke events using the model wi th ECG-LVH

<2.5% to <2.5%: 4,495 (97.6) <2.5% to 2.5-5.0%: 70 (4.0) <2.5% to >5.0%: 0 (0.0) Total: 4,565 (56.0)

2.5–5.0% to <2.5%: 110 (2.4) 2.5–5.0% to 2.5-5.0%: 1,551 (88.3) 2.5–5.0% to >5.0%: 93 (5.2) Total: 1,754 (21.5)

>5.0% to <2.5%: 0 (0.0) >5.0% to 2.5-5.0%: 135 (7.7) >5.0% to >5.0%: 1,702 (94.8) Total: 1,837 (22.5)

The model w ith and w ithout ECG-LV H and/or QTc interval Categorical NRI=0.035, p<0.001

Reclassification of subjects with stroke events using the model wi th ECG-LVH and QTc interval

<2.5% to <2.5%: 40 (87.0) <2.5% to 2.5–5.0%: 3 (5.5) <2.5% to >5.0%: 0 (0.0)

Total: 43 (14.5) 2.5–5.0% to <2.5%: 6 (13.0) 2.5–5.0% to 2.5–5.0%: 44 (80.0) 2.5–5.0% to >5.0%: 16 (8.2) Total: 66 (22.2)

Screening for CVD Ri sk Wi th ECG 142 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Ishikaw a, >5.0% to <2.5%: 0 (0.0) 201550 >5.0% to 2.5–5.0%: 8 (14.5) Fair >5.0% to >5.0%: 180 (91.8) (continued) Total: 188 (63.3)

Reclassification of subjects without stroke events using the model wi th ECG-LVH and QTc interval

<2.5% to <2.5%: 4403 (94.8) <2.5% to 2.5–5.0%: 157 (9.0) <2.5% to >5.0%: 5 (0.3) Total: 4,565 (56.0)

2.5–5.0% to <2.5%: 242 (5.2) 2.5–5.0% to 2.5–5.0%: 1,316 (75.9) 2.5–5.0% to >5.0%: 196 (11.0) Total: 1,754 (21.5)

>5.0% to <2.5%: 0 (0.0) >5.0% to 2.5–5.0%: 262 (15.1) >5.0% to >5.0%: 1,575 (88.7) Total: 1,837 (22.5)

Screening for CVD Ri sk Wi th ECG 143 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Fatal CV D Fatal CVD events: C-statistic Conventional Any ECG changes NR Any ECG changes 201452 events 2,236 (32.0) Continuous NRI risk factors C-Index for Fatal Continuous NRI for fatal CVD events Fair Categorical NRI CV D Ev ents Conventional risk factors and ECG Fatal or Fatal or nonfatal Conventional Conventional risk changes as present or not: 42.3 Nonfatal CVD events: 3,849 risk factors and factors: 0.705 (34.7–50.0) CVD events (55.0) major/minor (0.687–0.723) p<0.001 (combined) ECG changes Conventional risk Conventional risk factors and ECG All-cause mortality: factors and ECG changes w ith increasing severity: 42.3 All-cause 5,626 (80.5) changes: 0.719 (34.7–50.0) mortality (0.702–0.737) p<0.001 p<0.001 For sample w ith Conventional risk Categorical NRI for fatal CVD events ≥10 years follow up factors and ECG Conventional risk factors and ECG (used to calculate changes w ith changes as present or not: 7.1 (3.6– discrimination and increasing severity: 10.6) reclassification 0.719 (0.702– p<0.001 outcomes): 0.736) Conventional risk factors and ECG Fatal CVD events: p<0.001 changes w ith increasing severity: 7.2 837 (17) (3.7–10.7) Fatal or nonfatal C-Index for Fatal or p<0.001 CVD events: 2,092 Nonfatal CVD (38.6) Events Continuous NRI for Fatal or Nonfatal All-cause mortality: Conventional risk CV D Ev ents 2,225 (32.2) factors: 0.651 Conventional risk factors and ECG (0.639–0.663) changes as present or not: 29.1 Conventional risk (23.6–34.7) factors and ECG p<0.001 changes as present Conventional risk factors and ECG or not: 0.660 changes w ith increasing severity:29.2 (0.648–0.672) (23.7–34.8) p<0.001 p<0.001 Conventional risk factors and ECG Categorical NRI for Fatal or Nonfatal changes w ith CV D Ev ents increasing severity: Conventional risk factors and ECG 0.660 (0.648– changes as present or not: 3.8 (1.4– 0.671) 6.3) p<0.001 p<0.001

C-Index for all- cause mortality Conventional risk factors: 0.652 (0.640–0.664)

Screening for CVD Ri sk Wi th ECG 144 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Conventional risk Conventional risk factors and ECG 201452 factors and ECG changes w ith increasing severity: 4.2 Fair changes: 0.656 (1.8–6.7) (continued) (0.645–0.668) p<0.001 p<0.001 Conventional risk Continuous NRI for All-Cause Mortality factors and ECG Conventional Risk Factors and ECG changes w ith Changes as Present or Not: 22.7 increasing severity: (17.5–27.8) 0.656 (0.645– p<0.001 0.668) p<0.001 Categorical NRI for All-Cause Mortality Conventional Risk Factors and ECG C-Index for Fatal Changes as Present or Not: 1.9 (0.1– CVD Events (From 3.6) Table 3, model p<0.001 validation) Conventional risk Conventional Risk Factors and ECG factors: 0.705 Changes for Fatal CVD Events (From (0.703–0.707) Table 3, validation) Adjusted for Continuous NRI: 42.3 (42.0–42.4) : 0.706 Adjusted for optimism: 42.3 Conventional risk p<0.001 factors and ECG Categorical NRI: 7.1 (6.7–9.0) changes: 0.719 Adjusted for optimism: 8.6 (0.717–0.721) p<0.001 Adjusted for optimis m: 0.720 Conventional Risk Factors and ECG p<0.001 Changes for Fatal or Nonfatal CVD Events (From Table 3, validation) Fatal or Nonfatal Continuous NRI: 29.2 (28.4–29.2) CVD Events (From Table 3, validation) Conventional risk factors: 0.651 (0.649–0.653) Adjusted for optimism: 0.652 Conventional risk factors and ECG changes: 0.660 (0.658–0.662)

Screening for CVD Ri sk Wi th ECG 145 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Adjusted for Adjusted for optimism: 29.2 201452 optimism: 0.660 p<0.001 Fair p<0.001 Categorical NRI: 4.2 (3.5–5.6) (continued) Adjusted for optimism: 4.7 T w ave changes p<0.001 C-Index for Fatal CV D Ev ents T w ave changes Conventional risk Continuous NRI for Fatal CVD Events factors: 0.705 Conventional Risk Factors and ECG (0.687–0.723) Changes as Present or Not: 29.2 Conventional risk (21.5–36.8) factors and ECG p<0.001 change: 0.716 (0.699–0.734) Categorical NRI for Fatal CVD Events p<0.001 Conventional Risk Factors and ECG Changes as Present or Not: 5.4 (2.2– C-Index for Fatal 8.6) and Nonfatal CV D p<0.01 Events Conventional risk Continuous NRI for Fatal and Nonfatal factors: 0.651 CV D Ev ents (0.639–0.663) Conventional risk factors and ECG Conventional risk changes as present or not: 20.3 factors and ECG (14.7–25.9) change: 0.658 p<0.001 (0.647–0.670) p<0.001 Categorical NRI for Fatal and Nonfatal CV D Ev ents C-Index for All- Conventional risk factors and ECG Cause Mortality changes as present or not: 2.7 (0.6– Conventional risk 4.8) factors: 0.652 p<0.05 (0.640–0.664) Conventional risk factors and ECG change: 0.656 (0.644–0.668) p<0.01

Screening for CVD Ri sk Wi th ECG 146 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Ventricular Continuous NRI for All-Cause Mortality 201452 conduction delay Conventional risk factors and ECG Fair C-Index for Fatal changes as present or not: 16.5 (continued) CV D Ev ents (11.4–21.7) Conventional risk p<0.001 factors: 0.705 (0.687–0.723) Categorical NRI for All-Cause Mortality Conventional risk Conventional risk factors and ECG factors and ECG changes as present or not: 1.3 (-0.3– change: 0.708 3.0) (0.690–0.726) p>0.05 p>0.05

C-Index for Fatal Ventricular conduction delay and Nonfatal CV D Continuous NRI for Fatal CVD Events Events Conventional risk factors and ECG Conventional risk changes as present or not: 2.8 factors: 0.651 (-4.9–10.4) (0.639–0.663) p>0.05 Conventional risk factors and ECG Categorical NRI for Fatal CVD Events change: 0.655 Conventional risk factors and ECG (0.643–0.667) changes as present or not: 1.1 (0.1– p>0.05 2.1) p<0.05 C-Index for All- Cause Mortality Continuous NRI for Fatal and Nonfatal Conventional risk CV D Ev ents factors: 0.652 Conventional risk factors and ECG (0.640–0.664) changes as present or not: 5.5 Conventional risk (-0.1–11.1) factors and ECG p>0.05 change: 0.653 (0.642–0.665) p>0.05

LVH C-Index for Fatal CV D Ev ents

Screening for CVD Ri sk Wi th ECG 147 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Conventional risk Categorical NRI for Fatal and Nonfatal 201452 factors: 0.705 CV D Ev ents Fair (0.687–0.723) Conventional risk factors and ECG (continued) changes as present or not: 0.0 (-1.1– Conventional risk 1.2) factors and ECG p>0.05 change: 0.706 (0.688–0.724) Continuous NRI for All-Cause Mortality p>0.05 Conventional risk factors and ECG changes as present or not: 3.2 C-Index for Fatal (-2.0–8.4) and Nonfatal CV D p>0.05 Events Conventional risk Categorical NRI for All-Cause Mortality factors: 0.651 Conventional risk factors and ECG (0.639–0.663) changes as present or not: Conventional risk 0.2 (-0.5–1.0) factors and ECG p>0.05 change: 0.651 (0.639–0.663) LVH p>0.05 Continuous NRI for Fatal CVD Events Conventional risk factors and ECG C-Index for All- changes as present or not: 12.1 Cause Mortality (4.5–19.8) Conventional risk p<0.01 factors: 0.652 (0.640–0.664) Categorical NRI for Fatal CVD Events Conventional risk Conventional risk factors and ECG factors and ECG changes as present or not: 2.7 (1.0– change: 0.653 4.4) (0.641–0.665) p<0.01 p>0.05

Q w aves C-Index for Fatal CV D Ev ents Conventional risk factors: 0.705 (0.687–0.723)

Screening for CVD Ri sk Wi th ECG 148 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Conventional risk Continuous NRI for Fatal and Nonfatal 201452 factors and ECG CV D Ev ents Fair change: 0.709 Conventional risk factors and ECG (continued) (0.691–0.727) changes as present or not: 6.7 p<0.05 (1.1–12.3) p<0.05 C-Index for Fatal and Nonfatal CV D Categorical NRI for Fatal and Nonfatal Events CV D Ev ents Conventional risk Conventional risk factors and ECG factors: 0.651 changes as present or not: -1.1 (-2.3– (0.639–0.663) 0.1) Conventional risk p>0.05 factors and ECG change: 0.655 Continuous NRI for All-Cause Mortality (0.643–0.667) Conventional risk factors and ECG p<0.01 changes as present or not: 7.0 (1.8–12.1) C-Index for All- p<0.01 Cause Mortality Conventional risk Categorical NRI for All-Cause Mortality factors: 0.652 Conventional risk factors and ECG (0.640–0.664) changes as present or not: 0.7 (-0.2– Conventional risk 1.7) factors and ECG p>0.05 change: 0.654 (0.643–0.666) Q w aves p<0.05 Continuous NRI for Fatal CVD Events Conventional risk factors and ECG ST depressions changes as present or not: 5.3 C-Index for Fatal (-0.02–12.9) CV D Ev ents p>0.05 Conventional risk factors: 0.705 (0.687–0.723) Conventional risk factors and ECG change: 0.714 (0.697–0.732) p<0.001

Screening for CVD Ri sk Wi th ECG 149 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, C-Index for Fatal Categorical NRI for Fatal CVD Events 201452 and Nonfatal CV D Conventional risk factors and ECG Fair Events changes as present or not: 1.9 (0.7– (continued) Conventional risk 3.1) factors: 0.651 p<0.01 (0.639–0.663) Continuous NRI for Fatal and Nonfatal Conventional risk CV D Ev ents factors and ECG Conventional risk factors and ECG change: 0.660 changes as present or not: 3.9 (0.648–0.672) (-1.6–9.5) p<0.001 p>0.05

C-Index for All- Categorical NRI for Fatal and Nonfatal Cause Mortality CV D Ev ents Conventional risk Conventional risk factors and ECG factors: 0.652 changes as present or not: 0.7 (-0.2– (0.640–0.664) 1.8) Conventional risk p>0.05 factors and ECG change: 0.656 Continuous NRI for All-Cause Mortality (0.644–0.667) Conventional risk factors and ECG p<0.01 changes as present or not: 4.7 (-0.4–9.9) Resting heart Rate p>0.05 C-Index for Fatal CV D Ev ents Categorical NRI for All-Cause Mortality Conventional risk Conventional risk factors and ECG factors: 0.705 changes as present or not: 0.7 (0.0– (0.687–0.723) 1.4) Conventional risk p<0.05 factors and ECG change: 0.709 (0.691–0.727) p<0.001

C-Index for Fatal and Nonfatal CV D Events Conventional risk factors: 0.651 (0.639–0.663)

Screening for CVD Ri sk Wi th ECG 150 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Conventional risk ST depressions 201452 factors and ECG Continuous NRI for Fatal CVD Events Fair change: 0.652 Conventional risk factors and ECG (continued) (0.640–0.664) changes as present or not: 18.0 p>0.05 (10.4–25.6) p<0.001 C-Index for All- Cause Mortality Categorical NRI for Fatal CVD Events Conventional risk Conventional risk factors and ECG factors: 0.652 changes as present or not: 3.1 (0.7– (0.640–0.664) 5.4) Conventional risk p<0.01 factors and ECG change: 0.661 Continuous NRI for Fatal and Nonfatal (0.649–0.672) CV D Ev ents p<0.001 Conventional risk factors and ECG changes as present or not: 14.7 (9.1–20.3) p<0.001

Categorical NRI for Fatal and Nonfatal CV D Ev ents Conventional risk factors and ECG changes as present or not: 2.2 (0.4– 4.1) p<0.01

Continuous NRI for All-Cause Mortality Conventional risk factors and ECG changes as present or not: 11.1 (6.0–16.3) p<0.001

Screening for CVD Ri sk Wi th ECG 151 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Jorgensen, Categorical NRI for All-Cause Mortality 201452 Conventional risk factors and ECG Fair changes as present or not: 1.5 (0.3– (continued) 2.8) p<0.01

Resting heart Rate Continuous NRI for Fatal CVD Events Conventional risk factors and ECG changes as present or not: 14.1 (6.4–21.7) p<0.001

Categorical NRI for Fatal CVD Events Conventional risk factors and ECG changes as present or not: 0.9 (-1.8– 3.7) p>0.05

Continuous NRI for Fatal and Nonfatal CV D Ev ents Conventional risk factors and ECG changes as present or not: 7.3 (1.8–12.9) p<0.05

Categorical NRI for Fatal and Nonfatal CV D Ev ents Conventional risk factors and ECG changes as present or not: -0.2 (-1.4– 1.0) p>0.05

Continuous NRI for All-Cause Mortality Conventional risk factors and ECG changes as present or not: 21.4 (16.2–26.6) p<0.001

Categorical NRI for All-Cause Mortality Conventional risk factors and ECG changes as present or not: 3.7 (1.6– 5.7) p<0.001

Screening for CVD Ri sk Wi th ECG 152 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Shah, 201657 Primary: Derivation cohort: C-statistic NHA NES ECG C-statistic (95% CI) Hosmer-Lemeshow NRI, Total (Event NRI; Nonevent NRI), Fair CVD death CVD death: 574 IDI risk score chi-square values % (15.8) Calibration model NHANES ECG risk FRS model Secondary (only for base score model NHA NES ECG ris k Categorical NRI outcomes: Validation cohort: models) FRS model Fatal IHD: 0.80 score model Fatal IHD: 24 (17; 7) 10-year CV D death: 282 Reclassification (0.77–0.83) IHD Death: 185 Fatal CV D: 25 (12; 13) ischemic (4.4) NRI FRS model + Fatal CV D: 0.79 CV D Death: 281 All-cause death: 30 (11; 19) heart Fatal IHD: 166 ECG r is k (0.76–0.81) Death: 600 Continuous NRI disease (2.6) equation All-cause death: Fatal IHD: 57 (22; 35) (IHD) death All-cause death: 0.75 (0.73–0.77) FRS model Fatal CV D: 56 (21; 35) and all- 810 (12.8) PCE model IHD Death: 175 All-cause death: 53 (20; 33) cause death FRS model CV D Death: 237 PCE model + Fatal IHD: 0.79 Death: 458 A CC-AHA pooled cohort equation ECG r is k (0.76–0.82) model equation Fatal CVD: 0.76 Pooled cohort model Categorical NRI (0.73–0.78) (A CC-AHA) Fatal IHD: 14 (9; 5) All-cause death: IHD Death: 152 Fatal CV D: 25 (11; 14) 0.71 (0.69–0.73) CV D Death: 222 All-cause death: 19 (7; 12) Death: 447

Screening for CVD Ri sk Wi th ECG 153 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Shah, 201657 FRS variables FRS model + ECG Calibration statistics Continuous NRI Fair model risk equation f or the NHA NES Fatal IHD: 41 (17; 24) (continued) Fatal IHD: 0.82 ECG risk score for Fatal CV D: 54 (20; 34) FRS variables (0.79–0.85) CVD death w ere All-cause death: 35 (18; 17) model + ECG Fatal CV D: 0.80 adequate w ith risk equation (0.77–0.82) p=0.08 in the Framingham variables model All-cause death: derivation cohort Categorical NRI 0.75 (0.74–0.77) and p=0.22 in the Fatal IHD: 4 (3; 1) validation cohort. Fatal CV D: 11 (7; 4) Difference between All-cause death: 10 (6; 4) FRS model and Continuous NRI FRS model + ECG Fatal IHD: 37 (9; 28) risk equation Fatal CV D: 35 (7; 28) Fatal IHD: 0.03 All-cause death: 33 (8; 25) (0.01–0.05) Fatal CV D: 0.04 IDI, % (0.02–0.06) FRS model All-cause death: Absolute IDI 0.04 (0.03–0.05) Fatal IHD: 1.0 Fatal CV D: 1.6 PCE model All-cause death: 2.6 Fatal IHD: 0.80 Relative IDI (0.77–0.83) Fatal IHD: 25 Fatal CV D: 0.76 Fatal CV D: 35 (0.73–0.78) All-cause death: 38 All-cause death: 0.73 (0.71–0.75) A CC-AHA pooled cohort equation model PCE model + ECG Absolute IDI risk equation Fatal IHD: 0.7 Fatal IHD: 0.82 Fatal CV D: 2.0 (0.79–0.84) All-cause death: 1.9 Fatal CVD: 0.80 Relative IDI (0.78–0.83) Fatal IHD: 19 All-cause death: Fatal CV D: 47 0.76 (0.74–0.77) All-cause death: 25

Screening for CVD Ri sk Wi th ECG 154 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Shah, 201657 Difference between Framingham variables model Fair PCE model and Absolute IDI (continued) PCE model + ECG Fatal IHD: 0.2 risk equation Fatal CV D: 0.8 Fatal IHD: 0.02 All-cause death: 2.0 (0.01–0.03) Relative IDI Fatal CV D: 0.04 Fatal IHD: 7 (0.03–0.06) Fatal CV D: 13 All-cause death: All-cause death: 8 0.03 (0.02–0.03) 19

FRS variables model Fatal IHD: 0.83 (0.81–0.85) Fatal CVD: 0.81 (0.79–0.84) All-cause death: 0.78 (0.76–0.80)

FRS variables model + ECG risk equation Fatal IHD: 0.84 (0.82–0.87) Fatal CVD: 0.82 (0.80–0.85) All-cause death: 0.79 (0.77–0.82)

Difference between Framingham variables model and Framingham variables model + ECG risk equation Fatal IHD: 0.01 (0.01–0.02) Fatal CV D: 0.01 (0.01–0.02) All-cause death: 0.01 (0.01–0.02)

Screening for CVD Ri sk Wi th ECG 155 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Tereshchenko Sudden SCD: 311 (2.0) Reclassification Modified NR NR DTNPV discrimination ability 201451 cardiac NRI FRSi+DT NPV 1 NRI es timate=0.028, p=0.06 Fair death vs. modified Event NRI: 0.028 (2.8%) FRS Nonevent NRI: 0.0002 (0.02%)

Appropriately reclassified participants w ith SCD outcome into the higher risk categories: 3.4% appropriately reclassified into a higher risk group, 0.3% reclassified into a higher risk group inappropriately.

Reclassification w ith addition of DTNPV 1 for those w ith SCD events <5%, no change:135 <5% to 5–20%: 3 (2.2%) <5% to >20%: 0 Total: 138

5%–20% to <5%:1 (2.6%) 5%–20%, no change: 34 5%–20% to >20%: 3 (7.8%) Total: 38

>20% to <5%:0 >20% to 5-20%: 0 >20%, no change: 3 Total: 3

Classification based on modified FRS+ DTNPV1, Total <5%:136 5–20%: 37 >20%: 6 Total: 179

Reclassification for those w ithout SCD events <5%, no change <5%: 12,298 <5% to 5–20%: 27 (0.22%) <5% to >20%: 0 Total: 12,325

Screening for CVD Ri sk Wi th ECG 156 RTI–UNC EPC Appendix E Ta ble 2. Results of Included Studies for KQ 2 That Evaluated Resting ECG

Fir s t Author, N (%) With Outcome Year; Quality Outcome(s) Outcome(s) Measures Models Discrimination Calibration Reclassification Tereshchenko, 5%–20% to <5%:35 (6.6%) 201451 5%–20%, no change: 485 Fair 5%–20% to >20%: 7 (1.3%) (continued) Total: 527

>20% to <5%:0 >20% to 5–20%: 1 (5.6%) >20%, no change: 17 Total: 18

Classification based on modified FRS+ DTNPV1, Total <5%:12333 5–20%: 513 >20%: 24 Total: 12,870 a Adjudicated CHD events including hard CHD Events (acute MI and CHD deaths) and soft CHD events (hospitalization for angina and coronary revascularization). b FRS variables were age, sex, total and high-density lipoprotein cholesterol, systolic blood pressure, smoking. c FRS model included age, sex, systolic blood pressure, smoking history, serum cholesterol level, and serum high density lipoprotein level. d CHD defined as acute MI necessitating overnight hospitalizations, silent MI identified on serial ECGs, or death due to CHD. e CVD end points included CHD (CHD death and nonfatal MI), coronary artery bypass graft surgery/percutaneous transluminal coronary angioplasty and stroke. f A CHD event was defined as a validated definite or probable hospitalized myocardial infarction, a definite CHD death, an unrecognized myocardial infarction defined by ARIC ECG readings, or coronary revascularization. g Model included age, race, total & HDL cholesterol, systolic blood pressure, use of antihypertensive medication, and smoking status h Model included age, sex, body mass index, current smoking, alcohol intake >20 g/d, systolic blood pressure, antihypertensive medication use, diabetes mellitus, hyperlipidemia, and heart rate. i Modified FRS in this used the CHD Framingham risk score with age, gender, SBP, DM, HDL and total cholesterol, smoking, and BP-lowering therapy; they note that calculated FRS scores were not directly used due to possible issues of the applicability to different ethnic groups and were adjusted for race

Abbreviations: ACC=American College of Cardiology AHA=American Heart Association; AUC=area under the curve; AUROC=area under the receiver operating characteristic curve; CHD=coronary heart disease; CI=confidence interval; CV=cardiovascular; CVD=cardiovascular disease; CVRF=cardiovascular risk factors; ECG=electrocardiogram; DTNPVI=Deep Terminal Negativity of the P Wave in V1 ; FRS=Framingham Risk Score; IDI=integrated discrimination improvement; IHD=ischemic heart disease; KQ=key question; LVH= left ventricular hypertrophy; MI=myocardial infarction; NA=not applicable; NHANES=National Health and Nutrition Examination Survey; NRI=net reclassification improvement; PCE=pooled cohort equation; QTc=corrected QT int erval; SCD=sudden cardiac death.

Screening for CVD Ri sk Wi th ECG 157 RTI–UNC EPC Appendix E Ta ble 3. Number and Percentage of Execise ECGs With Abnormalities in Studies Included for KQ 2

First Author, Year Sam ple Size n (%) with Abnormalities ECG Findings Evaluated Source of Patients Country Aktas, 200459 3,554 371 (10.4%) had ischemic Exercise ECG according to Bruce (or Consecutive participants U.S. ST-segment changes modified Bruce) protocol; ischemic ST presenting for an executive 549 (15.4%) had abnormal abnormality using a 12-lead, symptom- physical. Self-referred. based on functional limited exercise ECG capacity or HR recovery Chang, 201545 988 (946 w ith 116/946 (12.3%) w ith Exercise ECG according to Bruce People w ho had both CACS and U.S. follow up) follow up had ischemic protocol; stress-induced ischemia stress SPECT for clinically exercise ECG; 75% w ith identified via ECG during symptom- indicated reasons at the Heart and >8 METs ; 22% 5-8 METs ; limited exercise treadmill testing; METs Vascular Center <5 METs and DTS Cournot, 200660 1,051 89 (8.5%) had positive Symptom-limited exercise ECG Consecutive asymptomatic people France exercise test self-referred or referred by PCPs and cardiologists for evaluation of risk factors and routine screening Cournot, 200956 2,709 163 (6.4%) had positive A positive exercise testg during a Apparently healthy asymptomatic France exercise test symptom-limited exercise ECG w ith people self-referred (20%) or orthogonal and V1 to V6 leads referred by PCPs (27%) or other providers to a preventive cardiology unit Erikssen, 200462 Assessment 1 (1972– 205 (10.2%) positive Resting ECG and a symptom-limited Apparently healthy males ages 40– Norw ay 1975): 2,014 exercise test bicycle exercise ECG test 60 years recruited from five Assessment 2 (1980– governmental agencies w ho 1982): 1,428 238 (16.7%) positive participated in a cardiovascular risk exercise test assessment Abbreviations: CACS=coronary artery calcium score; DT S=Duke treadmill score; ECG=electrocardiogram; HR=hazard ratio; MET s=metabolic equivalents; PCPs=primary care physicians; SPECT=Single Photon Emission Computed Tomography; U.S.=United States.

Screening for CVD Ri sk Wi th ECG 158 RTI–UNC EPC Appendix E Ta ble 4. Number and Percentage of Resting ECGs With Abnormalities in Studies Included for KQ 2

First Author, Year Sam ple Size n (%) With Abnormalities ECG Findings Evaluated Source of Patients Country Auer et al, 201255 2,192 782 (36%) any abnormality Majora and minorb 12-lead Population-based cohort assessing U.S. 506 (23%) major ECG abnormalities classified body composition, long-ter m 276 (13%) minor using the Minnesota Coding conditions, and incident mobility 1,410 (64%) no abnormality System limitation in an older adult cohort (1997–1998) Badheka et al, 201354 6,025 3,291 (54.6%) had any ECG Major and Minor 12-lead Population-based survey to collect U.S. abnormalities ECG abnormalities classified information on the health and using Minnesota Coded nutrition of U.S. households (1988– 1994) Badheka et al, 201353 7,928 1,919 (24.2%) ST-segment 12-lead ECG ST-T w ave Population-based survey to collect U.S. elevation (>8 mV) in aVR; abnormalities in lead aVR information on the health and 1,459 (18.4%) T-w ave classified by the Minnesota nutrition of U.S. households (1988– amplitude 0.1 mV or greater in Code 1994) aVR Denes, 200758 1,264c Data only reported for the larger Major,e minor,f and incidentg Population-based study on common U.S. WHI sample of 14,749: 12-lead ECG changes using causes of morbidity/ mortality among 910 (6.2%) major the Novacode criteria postmenopausal w omen (1993– 4,095 (27.8%) minor 1998) 9,744 (66.1%) none Folsom, 200363 14,054 NR LVH using a 12-lead ECG Population-based study of 4 U.S. U.S. and the Cornell score communities (1987–1989) Ishikaw a, 201550 10,643 162 (1.5%) had prolonged QTc Prolonged corrected QT Government-sponsored screening to Japan intervals (QTc) intervalsi and LVHj on clarify the risk factors for cardio/ 12-lead ECG cerebrovascular diseases in the general population (1992-1995) Jorgensen, 201452 6,991g 2,140 (30.6%) any ECG Major and Minor 12-lead The Copenhagen City Heart Study Denmark abnormalities ECG abnormalities classified (1976–1978) 1,163 (16.6%) major using Minnesota Code; also 353 (5.0%) intermediate reported outcomes for some 624 (8.9%) minor single ECG changesm Shah, 201657 9,969 NR (reported mean and SD for ECG Risk Score including Population-based survey to collect U.S. (derivation: 3,640, the variables in the ECG Risk frontal T axis, corrected QT information on health and nutrition; validation: Score) interval, T axis, heart rate, NHA NES I (1971–1975) and 6,329) age, sex, age*sex interaction NHA NES III (1988–1994) term (selected from majoro and minorp abnormalities) Tereshchenko, 201451 15,375k 167 (1.1%) had the specific Resting 12-lead, P w ave Population-based study of 4 U.S. U.S. DTNPV1 abnormality morphology (specifically communities (1987–1989) DTNPV 1r) Abbreviations: DTNPVI=Deep Terminal Negativity of the P Wave in V1; ECG=electrocardiogram; LVH=left ventricular hypertrophy; NHANES=National Health and Nutrition Examination Survey; NR=not reported; QTc=corrected QT interval; SD=standard deviation; U.S.=United States; WHI=Women’s Health Initiative.

Screening for CVD Ri sk Wi th ECG 159 RTI–UNC EPC Appendix E Ta ble 5. Results of Included Randomized, Controlled Trials Reporting Harms (KQ 3)

CV Events Mortality Due to Due to Labeling Har m s of Subsequent Procedures/ Fir s t Screening Arrhythmia Screening Injuries Anxiety G1 N (%) Interventions Author, G1 N (%) G1 N (%) G1 N (%) G1 N (%) G1 N (%) G2 N (%) G1 N (%) Year G1 (N) G2 N (%) G2 N (%) G2 N (%) G2 N (%) G2 N (%) HR (95% G2 N (%) Trial Name G2 (N) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) HR (95% CI) CI) HR (95% CI) Lievre et al, Screened NR (all-cause NR NR NR NR NR NR (number of revascularizations 201147 (316) mortality described reported in KQ 1, but NR w hether any DY NA MIT Not in KQ 1) resulted in harms; 18 vs. 21) screened (315) Turrini et Screened NR (all-cause NR NR NR NR NR 20/262 (7.6%) patients w ho underw ent al, 201548 (262) mortality described ETT had a positive result. Of those, 17 DA DDY -D Not in KQ 1) underw ent coronary angiography. It w as screened positive for critical stenosis in 12/17 (258) (70.6%), and all 12 underw ent revascularization procedures (7 percutaneous and 5 surgery). One patient having percutaneous revascularization had an acute MI (nonfatal) 3 days after the procedure. Abbreviations: CI=confidence interval; CV=cardiovascular; DADDY-D=Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients; DYNAMIT=Do You Need to Assess Myocardial Ischemia in Type-2 diabetes; ETT=exercise treadmill test; G=group; HR=hazard ratio; KQ=key question; MI=myocardial infarction; N=sample size; NR=not reported.

Screening for CVD Ri sk Wi th ECG 160 RTI–UNC EPC References

1. Chou R, High Value Care Task Force of the American College of P. Cardiac screening with electrocardiography, stress echocardiography, or myocardial perfusion imaging: advice for high-value care from the American College of Physicians. Ann Intern Med. 2015 Mar 17;162(6):438-47. doi: 10.7326/M14- 1225. PMID: 25775317. 2. Wolk MJ, Bailey SR, Doherty JU, et al. ACCF/AHA/ASE/ASNC/HFSA/HRS/SCAI/SCCT/SCMR/STS 2013 multimodality appropriate use criteria for the detection and risk assessment of stable ischemic heart disease: a report of the American College of Cardiology Foundation Appropriate Use Criteria Task Force, American Heart Association, American Society of Echocardiography, American Society of Nuclear Cardiology, Heart Failure Society of America, Heart Rhythm Society, Society for Cardiovascular Angiography and Interventions, Society of Cardiovascular Computed Tomography, Society for Cardiovascular Magnetic Resonance, and Society of Thoracic Surgeons. J Am Coll Cardiol. 2014 Feb 4;63(4):380-406. doi: 10.1016/j.jacc.2013.11.009. PMID: 24355759. 3. American Academy of Family Physicians. Annual EKGs for Low-risk Patients. 2012. http://www.aafp.org/patient-care/clinical-recommendations/all/cw-ekg.html. Accessed 18 Nov, 2015. 4. Lim LS, Haq N, Mahmood S, et al. Atherosclerotic cardiovascular disease screening in adults: American College of Preventive Medicine position statement on preventive practice. Am J Prev Med. 2011 Mar;40(3):381 e1-10. doi: 10.1016/j.amepre.2010.11.021. PMID: 21335273. 5. Greenland P, Alpert JS, Beller GA, et al. 2010 ACCF/AHA guideline for assessment of cardiovascular risk in asymptomatic adults: a report of the American College of Cardiology Foundation/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2010 Dec 14;56(25):e50-103. doi: 10.1016/j.jacc.2010.09.001. PMID: 21144964. 6. Lauer M, Froelicher ES, Williams M, et al. Exercise testing in asymptomatic adults: a statement for professionals from the American Heart Association Council on Clinical Cardiology, Subcommittee on Exercise, Cardiac Rehabilitation, and Prevention. Circulation. 2005 Aug 2;112(5):771-6. doi: 10.1161/CIRCULATIONAHA.105.166543. PMID: 15998671. 7. National Heart, Lung, and Blood Institute. 2013 Report on the Assessment of Cardiovascular Risk: Full Work Group Report Supplement: National Heart, Lung, and Blood Institute; 2013. 8. Goff DC, Jr., Lloyd-Jones DM, Bennett G, et al. 2013 ACC/AHA guideline on the assessment of cardiovascular risk: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. J Am Coll Cardiol. 2014 Jul 01;63(25 Pt B):2935-59. doi: 10.1016/j.jacc.2013.11.005. PMID: 24239921. 9. D'Agostino RB, Sr., Vasan RS, Pencina MJ, et al. General cardiovascular risk profile for use in primary care: the Framingham Heart Study. Circulation. 2008 Feb 12;117(6):743-53. doi: 10.1161/CIRCULATIONAHA.107.699579. PMID: 18212285. 10. Anderson TJ, Gregoire J, Hegele RA, et al. 2012 update of the Canadian Cardiovascular Society guidelines for the diagnosis and treatment of dyslipidemia for the prevention of cardiovascular disease in the adult. Can J Cardiol. 2013 Feb;29(2):151-67. doi: 10.1016/j.cjca.2012.11.032. PMID: 23351925. 11. Hippisley-Cox J, Coupland C, Vinogradova Y, et al. Predicting cardiovascular risk in England and Wales: prospective derivation and validation of QRISK2. BMJ. 2008 Jun 28;336(7659):1475-82. doi: 10.1136/bmj.39609.449676.25. PMID: 18573856. 12. National Institute for Health and Care Excellence. Lipid Modification: cardiovascular risk assessment and the modification of for the primary and secondary prevention of cardiovascular disease. United Kingdom: National Institute for Health and Care Excellence; 2014. 13. Ridker PM, Paynter NP, Rifai N, et al. C-reactive protein and parental history improve global cardiovascular risk prediction: the Reynolds Risk Score for men. Circulation. 2008 Nov 25;118(22):2243- 51, 4p following 51. doi: 10.1161/CIRCULATIONAHA.108.814251. PMID: 18997194. 14. Ridker PM, Buring JE, Rifai N, et al. Development and validation of improved algorithms for the assessment of global cardiovascular risk in women: the Reynolds Risk Score. JAMA. 2007 Feb 14;297(6):611-9. doi: 10.1001/jama.297.6.611. PMID: 17299196. 15. Woodward M, Brindle P, Tunstall-Pedoe H, et al. Adding social deprivation and family history to cardiovascular risk assessment: the ASSIGN score from the Scottish Heart Health Extended Cohort (SHHEC). Heart. 2007 Feb;93(2):172-6. doi: 10.1136/hrt.2006.108167. PMID: 17090561. 16. Scottish Intercollegiate Guidelines Network. Risk estimation and the prevention of cardiovascular disease: a national clinical guideline. Edinburgh: Scottish Intercollegiate Guidelines Network; 2007. 17. Chambless LE, Folsom AR, Sharrett AR, et al. Coronary heart disease risk prediction in the Atherosclerosis Risk in Communities (ARIC) study. J Clin Epidemiol. 2003 Sep;56(9):880-90. PMID: 14505774.

Screening for CVD Ri sk Wi th ECG 161 RTI–UNC EPC References

18. Conroy RM, Pyorala K, Fitzgerald AP, et al. Estimation of ten-year risk of fatal cardiovascular disease in Europe: the SCORE project. Eur Heart J. 2003 Jun;24(11):987-1003. PMID: 12788299. 19. Perk J, De Backer G, Gohlke H, et al. European Guidelines on cardiovascular disease prevention in clinical practice (version 2012). The Fifth Joint Task Force of the European Society of Cardiology and Other Societies on Cardiovascular Disease Prevention in Clinical Practice (constituted by representatives of nine societies and by invited experts). Eur Heart J. 2012 Jul;33(13):1635-701. doi: 10.1093/eurheartj/ehs092. PMID: 22555213. 20. Assmann G, Cullen P, Schulte H. Simple scoring scheme for calculating the risk of acute coronary events based on the 10-year follow-up of the prospective cardiovascular Munster (PROCAM) study. Circulation. 2002 Jan 22;105(3):310-5. PMID: 11804985. 21. National Cholesterol Education Program Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults. Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III) final report. Circulation. 2002 Dec 17;106(25):3143-421. PMID: 12485966. 22. Anderson KM, Odell PM, Wilson PW, et al. Cardiovascular disease risk profiles. Am Heart J. 1991 Jan;121(1 Pt 2):293-8. PMID: 1985385. 23. Anderson KM, Wilson PW, Odell PM, et al. An updated coronary risk profile. A statement for health professionals. Circulation. 1991 Jan;83(1):356-62. PMID: 1984895. 24. Wilson PW, D'Agostino RB, Levy D, et al. Prediction of coronary heart disease using risk factor categories. Circulation. 1998 May 12;97(18):1837-47. PMID: 9603539. 25. Wolf PA, D'Agostino RB, Belanger AJ, et al. Probability of stroke: a risk profile from the Framingham Study. Stroke. 1991 Mar;22(3):312-8. PMID: 2003301. 26. U. S. Preventive Services Task Force. Aspirin Use to Prevent Cardiovascular Disease and Colorectal Cancer: Preventive Medication. Rockville, MD: U. S. Preventive Services Task Force; November 2016. https://www.uspreventiveservicestaskforce.org/Page/Document/RecommendationStatementFinal/aspirin- to-prevent-cardiovascular-disease-and-cancer 27. Vandvik PO, Lincoff AM, Gore JM, et al. Primary and secondary prevention of cardiovascular disease: Antithrombotic Therapy and Prevention of Thrombosis, 9th ed: American College of Chest Physicians Evidence-Based Clinical Practice Guidelines. Chest. 2012 Feb;141(2 Suppl):e637S-68S. doi: 10.1378/chest.11-2306. PMID: 22315274. 28. Goldstein LB, Bushnell CD, Adams RJ, et al. Guidelines for the primary prevention of stroke: a guideline for healthcare professionals from the American Heart Association/American Stroke Association. Stroke. 2011 Feb;42(2):517-84. doi: 10.1161/STR.0b013e3181fcb238. PMID: 21127304. 29. Pignone M, Alberts MJ, Colwell JA, et al. Aspirin for primary prevention of cardiovascular events in people with diabetes: a position statement of the American Diabetes Association, a scientific statement of the American Heart Association, and an expert consensus document of the American College of Cardiology Foundation. Diabetes Care. 2010 Jun;33(6):1395-402. doi: 10.2337/dc10-0555. PMID: 20508233. 30. U. S. Preventive Services Task Force. Final Update Summary: Statin Use for the Primary Prevention of Cardiovascular Disease in Adults: Preventive Medication. Rockville, MD: U.S. Preventive Services Task Force; November 2016. https://www.uspreventiveservicestaskforce.org/Page/Document/UpdateSummaryFinal/statin-use-in-adults- preventive-medication1 31. Stone NJ, Robinson JG, Lichtenstein AH, et al. 2013 ACC/AHA guideline on the treatment of blood cholesterol to reduce atherosclerotic cardiovascular risk in adults: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines. Circulation. 2014 Jun 24;129(25 Suppl 2):S1-45. doi: 10.1161/01.cir.0000437738.63853.7a. PMID: 24222016. 32. National Institute for Health and Care Excellence. Cardiovascular disease: risk assessment and reduction, including lipid modification. United Kingdom: National Institute for Health and Care Excellence; 2016. https: //www.nice.org.uk/Guidance/CG181. Accessed May 30, 2017. 33. Mosca L, Benjamin EJ, Berra K, et al. Effectiveness-based guidelines for the prevention of cardiovascular disease in women--2011 update: a guideline from the american heart association. Circulation. 2011 Mar 22;123(11):1243-62. doi: 10.1161/CIR.0b013e31820faaf8. PMID: 21325087. 34. Siscovick DS, Barringer TA, Fretts AM, et al. Omega-3 Polyunsaturated Fatty Acid () Supplementation and the Prevention of Clinical Cardiovascular Disease: A Science Advisory From the American Heart Association. Circulation. 2017 Apr 11;135(15):e867-e84. doi: 10.1161/CIR.0000000000000482. PMID: 28289069.

Screening for CVD Ri sk Wi th ECG 162 RTI–UNC EPC References

35. National Institute for Health and Care Excellence. Do Not Do Recommendation. United Kingdom: National Institute for Health and Care Excellence; 2017. https://www.nice.org.uk/donotdo/do-not-offer- omega-3-fatty-acid-compounds-for-the-prevention-of-cvd-to-any-of-the-following-people-who-are-being- treated-for-primary-prevention-people-who-are-being-treated-for-secondary-prevention-people-with-ckd- people-with-type-1-diabetes-people-with-t. Accessed May 31, 2017. 36. George MG, Tong X, Sonnenfeld N, et al. Recommended use of aspirin and other antiplatelet medications among adults--National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey, United States, 2005-2008. MMWR Suppl. 2012 Jun 15;61(2):11-8. PMID: 22695458. 37. Mainous AG, Tanner RJ, Shorr RI, et al. Use of aspirin for primary and secondary cardiovascular disease prevention in the United States, 2011-2012. J Am Heart Assoc. 2014 Jul 14;3(4)doi: 10.1161/JAHA.114.000989. PMID: 25023071. 38. Malayala SV, Raza A. Compliance with USPSTF recommendations on aspirin for prevention of cardiovascular disease in men. Int J Clin Pract. 2016 Nov;70(11):898-906. doi: 10.1111/ijcp.12869. PMID: 27619938. 39. Fiscella K, Winters PC, Mendoza M, et al. Do clinicians recommend aspirin to patients for primary prevention of cardiovascular disease? J Gen Intern Med. 2015 Feb;30(2):155-60. doi: 10.1007/s11606-014- 2985-8. PMID: 25092016. 40. Kumar A, Fonarow GC, Eagle KA, et al. Regional and practice variation in adherence to guideline recommendations for secondary and primary prevention among outpatients with atherothrombosis or risk factors in the United States: a report from the REACH Registry. Crit Pathw Cardiol. 2009 Sep;8(3):104- 11. doi: 10.1097/HPC.0b013e3181b8395d. PMID: 19726929. 41. Noto TJ, Jr., Johnson LW, Krone R, et al. 1990: a report of the Registry of the Society for Cardiac Angiography and Interventions (SCA&I). Cathet Cardiovasc Diagn. 1991 Oct;24(2):75-83. PMID: 1742788. 42. Patel MR, Peterson ED, Dai D, et al. Low diagnostic yield of elective coronary angiography. N Engl J Med. 2010 Mar 11;362(10):886-95. doi: 10.1056/NEJMoa0907272. PMID: 20220183. 43. Schulman-Marcus J, Feldman DN, Rao SV, et al. Characteristics of Patients Undergoing Cardiac Catheterization Before Noncardiac Surgery: A Report From the National Cardiovascular Data Registry CathPCI Registry. JAMA Intern Med. 2016 May 01;176(5):611-8. doi: 10.1001/jamainternmed.2016.0259. PMID: 27018942. 44. McFalls EO, Ward HB, Moritz TE, et al. Coronary-artery revascularization before elective major vascular surgery. New England Journal of Medicine. 2004 Dec 30;351(27):2795-804. doi: DOI 10.1056/NEJMoa041905. PMID: WOS:000226004300004. 45. Chang SM, Nabi F, Xu J, et al. Value of CACS compared with ETT and myocardial perfusion imaging for predicting long-term cardiac outcome in asymptomatic and symptomatic patients at low risk for coronary disease: clinical implications in a multimodality imaging world. JACC Cardiovasc Imaging. 2015 Feb;8(2):134-44. doi: 10.1016/j.jcmg.2014.11.008. PMID: 25677886. 46. Harris RP, Helfand M, Woolf SH, et al. Current methods of the US Preventive Services Task Force: a review of the process. Am J Prev Med. 2001 Apr;20(3 Suppl):21-35. PMID: 11306229. 47. Lievre MM, Moulin P, Thivolet C, et al. Detection of silent myocardial ischemia in asymptomatic patients with diabetes: results of a randomized trial and meta-analysis assessing the effectiveness of systematic screening. Trials. 2011;12:23. doi: 10.1186/1745-6215-12-23. PMID: 21269454. 48. Turrini F, Scarlini S, Mannucci C, et al. Does coronary Atherosclerosis Deserve to be Diagnosed earlY in Diabetic patients? The DADDY-D trial. Screening diabetic patients for unknown coronary disease. Eur J Intern Med. 2015 Jul;26(6):407-13. doi: 10.1016/j.ejim.2015.05.006. PMID: 26058988. 49. Turrini F, Messora R, Giovanardi P, et al. Screening asymptomatic patients with diabetes for unknown coronary artery disease: does it reduce risk? An open-label randomized trial comparing a strategy based on exercise testing aimed at revascularization with management based on pharmacological/behavioural treatment of traditional risk factors. DADDY-D Trial (Does coronary Atherosclerosis Deserve to be Diagnosed and treated early in Diabetics?). Trials; 2009. p. 119. 50. Ishikawa J, Ishikawa S, Kario K. Prolonged corrected QT interval is predictive of future stroke events even in subjects without ECG-diagnosed left ventricular hypertrophy. Hypertension. 2015 Mar;65(3):554-60. doi: 10.1161/hypertensionaha.114.04722. PMID: 25534703. 51. Tereshchenko LG, Henrikson CA, Sotoodehnia N, et al. Electrocardiographic deep terminal negativity of the P wave in V(1) and risk of sudden cardiac death: the Atherosclerosis Risk in Communities (ARIC) study. J Am Heart Assoc. 2014 Dec;3(6):e001387. doi: 10.1161/jaha.114.001387. PMID: 25416036.

Screening for CVD Ri sk Wi th ECG 163 RTI–UNC EPC References

52. Jorgensen PG, Jensen JS, Marott JL, et al. Electrocardiographic changes improve risk prediction in asymptomatic persons age 65 years or above without cardiovascular disease. J Am Coll Cardiol. 2014 Sep 2;64(9):898-906. doi: 10.1016/j.jacc.2014.05.050. PMID: 25169175. 53. Badheka AO, Patel NJ, Grover PM, et al. ST-T wave abnormality in lead aVR and reclassification of cardiovascular risk (from the National Health and Nutrition Examination Survey-III). Am J Cardiol. 2013 Sep 15;112(6):805-10. doi: 10.1016/j.amjcard.2013.04.058. PMID: 23764245. 54. Badheka AO, Patel N, Tuliani TA, et al. Electrocardiographic abnormalities and reclassification of cardiovascular risk: insights from NHANES-III. Am J Med. 2013 Apr;126(4):319-26.e2. doi: 10.1016/j.amjmed.2012.10.020. PMID: 23415052. 55. Auer R, Bauer DC, Marques-Vidal P, et al. Association of major and minor ECG abnormalities with coronary heart disease events. JAMA. 2012 Apr 11;307(14):1497-505. doi: 10.1001/jama.2012.434. PMID: 22496264. 56. Cournot M, Taraszkiewicz D, Cambou JP, et al. Additional prognostic value of physical examination, exercise testing, and arterial ultrasonography for coronary risk assessment in primary prevention. Am Heart J. 2009 Nov;158(5):845-51. doi: 10.1016/j.ahj.2009.08.017. PMID: 19853707. 57. Shah AJ, Vaccarino V, Janssens AC, et al. An electrocardiogram-based risk equation for incident cardiovascular disease from the National Health and Nutrition Examination Survey. JAMA Cardiol. 2016 Aug 3doi: 10.1001/jamacardio.2016.2173. PMID: 27487404. 58. Denes P, Larson JC, Lloyd-Jones DM, et al. Major and minor ECG abnormalities in asymptomatic women and risk of cardiovascular events and mortality. JAMA. 2007 Mar 7;297(9):978-85. doi: 10.1001/jama.297.9.978. PMID: 17341712. 59. Aktas MK, Ozduran V, Pothier CE, et al. Global risk scores and exercise testing for predicting all-cause mortality in a preventive medicine program. JAMA. 2004 Sep 22;292(12):1462-8. doi: 10.1001/jama.292.12.1462. PMID: 15383517. 60. Cournot M, Taraszkiewicz D, Galinier M, et al. Is exercise testing useful to improve the prediction of coronary events in asymptomatic subjects? Eur J Cardiovasc Prev Rehabil. 2006 Feb;13(1):37-44. PMID: 16449862. 61. Strom Moller C, Zethelius B, Sundstrom J, et al. Persistent ischaemic ECG abnormalities on repeated ECG examination have important prognostic value for cardiovascular disease beyond established risk factors: a population-based study in middle-aged men with up to 32 years of follow-up. Heart. 2007 Sep;93(9):1104- 10. doi: 10.1136/hrt.2006.109116. PMID: 17483125. 62. Erikssen G, Bodegard J, Bjornholt JV, et al. Exercise testing of healthy men in a new perspective: from diagnosis to prognosis. Eur Heart J. 2004 Jun;25(11):978-86. doi: 10.1016/j.ehj.2004.04.009. PMID: 15172470. 63. Folsom AR, Chambless LE, Duncan BB, et al. Prediction of coronary heart disease in middle-aged adults with diabetes. Diabetes Care. 2003 Oct;26(10):2777-84. PMID: 14514579.

Screening for CVD Ri sk Wi th ECG 164 RTI–UNC EPC